- Email: [email protected]
Accommodation Closed-loop Control of Dieseline Fueled Flexible Fuel Engine Based on In-cylinder Pressure Sensor
Preprints, 8th IFAC International Symposium on Preprints, 8th IFAC International Symposium on Advances in Automotive Control Advances in Automotive Control Symposium Preprints, 8th IFAC International on Preprints, 8th IFACNorrköping, International Symposium ononline at www.sciencedirect.com June 19-23, 2016. Sweden Available June 19-23,in Norrköping, Sweden Advances in2016. Automotive Control Advances Automotive Control June June 19-23, 19-23, 2016. 2016. Norrköping, Norrköping, Sweden Sweden
ScienceDirect
IFAC-PapersOnLine 49-11 (2016) 202–209 Accommodation Accommodation Closed-loop Closed-loop Control Control of of Dieseline Dieseline Fueled Fueled Flexible Flexible Fuel Fuel Engine Engine Accommodation Control of Dieseline Fueled Flexible Fuel Engine Based on In-cylinder Pressure Sensor Accommodation Closed-loop Closed-loop Control of Dieseline Fueled Flexible Fuel Engine Based on In-cylinder Pressure Sensor Based on In-cylinder Pressure Sensor Based on In-cylinder Pressure Sensor Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Yang*.
Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Yang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Minggao Haiyan Huang*. Yaodong Jinli Hu*.Wang*. Tianyuan Zhou*. Ouyang*. Changsheng Yao*. Fuyuan Yang*. Yang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. *State *State Key Key Laboratory Laboratory of of Automotive Automotive Safety Safety and and Energy, Energy, Tsinghua Tsinghua University, University, Beijing Beijing 100084, 100084, China China (e-mail: [email protected]). *State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing (e-mail: *State Key Laboratory of Automotive Safety [email protected]). and Energy, Tsinghua University, Beijing 100084, 100084, China China (e-mail: (e-mail: [email protected]). [email protected]).
Abstract: Abstract: This This paper paper conducts conducts investigation investigation on on accommodation accommodation closed-loop closed-loop control control of of dieseline dieseline fueled fueled flexible fuel engine. In our previous work, gasoline/diesel blend ratio estimation method was proposed proposed Abstract: This paper investigation on control of fueled flexible fuel engine. our previous work, gasoline/diesel blend closed-loop ratio estimation method was Abstract: This paperInconducts conducts investigation on accommodation accommodation closed-loop control of dieseline dieseline fueled and acted actedfuel as aaengine. feedforward controller. In this thisgasoline/diesel paper, power power performance performance closed-loopmethod controlwas systems with flexible In previous blend proposed and as feedforward In paper, closed-loop control systems with flexible fuel engine. In our ourcontroller. previous work, work, gasoline/diesel blend ratio ratio estimation estimation method was proposed or without blend ratio estimation model are simulated respectively at constant and variable load to andwithout acted as asblend feedforward controller. In this this paper, powerrespectively performanceatclosed-loop closed-loop control systems with or ratio estimation model arepaper, simulated constant and variable load to and acted aa feedforward controller. In power performance control systems with demonstrate the necessity of feedforward controller to control system. Results show that, whether at or without blend ratio estimation model are simulated respectively at constant and variable load to demonstrate the necessity of feedforward controller to control system. that, whether at or without blend ratio estimation model are simulated respectively at Results constantshow and variable load to constant or variable load, power indicators of the system with feedforward controller track the target demonstrate the necessity of feedforward controller to control system. Results show that, whether at constant or variable load, power indicatorscontroller of the system with system. feedforward controller trackwhether the target demonstrate the necessity of feedforward to control Results show that, at value well. well. However, forpower system without offeedforward feedforward controller, power controller performance deteriorates constant or load, indicators with track the value However, for system without power performance constant or variable variable load, power indicators of the the system system controller, with feedforward feedforward controller trackdeteriorates the target target obviously, especially especially in in variable loadwithout condition. Besides, accommodation accommodation closed-loop controldeteriorates system for for value for feedforward controller, performance obviously, load condition. Besides, closed-loop control system value well. well. However, However, variable for system system without feedforward controller, power power performance deteriorates dieseline fueled flexible fuel engine is established. The algorithm is executed by adjusting the main obviously, especially in variable load condition. Besides, accommodation closed-loop control system for dieseline flexible fuel engine is established. The accommodation algorithm is executed by adjusting the main obviously,fueled especially in variable load condition. Besides, closed-loop control system for injection timing timing to seek seek the most most economy economyestablished. point on on the theThe premise that the the power performance of the the engine dieseline fueled algorithm is executed by the main injection point premise that performance of dieseline fueled toflexible flexiblethefuel fuel engine engine is is established. The algorithm is power executed by adjusting adjusting theengine main can be be ensured ensured simultaneously. Compared with open-loop open-loop control algorithm, closed-loop control system injection timing to economy on that the performance of engine can with control injection timingsimultaneously. to seek seek the the most mostCompared economy point point on the the premise premise thatalgorithm, the power powerclosed-loop performancecontrol of the the system engine is less sensitive to fuel estimation model error, which proves to be of better consistency. Moreover, the canless be sensitive ensured simultaneously. simultaneously. Compared with which open-loop control algorithm, closed-loop control control system is to fuel estimation model error, proves to bealgorithm, of better consistency. Moreover, the can be ensured Compared with open-loop control closed-loop system practical output torque follows the target value well and the indicated efficiency can also be optimized. is less sensitive to fuel estimation model error, which proves to be of better consistency. Moreover, the practical output torque the target andproves the indicated can also beMoreover, optimized.the is less sensitive to fuelfollows estimation modelvalue error,well which to be ofefficiency better consistency. practical output torque follows the target value well and the indicated efficiency can also be optimized. practical output torque follows the target value well and the indicated efficiency can also be optimized. © 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: in-cylinder pressure sensor; closed-loop combustion control; dieseline; flexible fuel engine; Keywords: in-cylinder pressure sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation control;pressure economy optimization; PPCI; Keywords: in-cylinder sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation control;pressure economy optimization; PPCI; Keywords: in-cylinder sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation accommodation control; control; economy economy optimization; optimization; PPCI; PPCI; that that fuel fuel with with characters characters between between diesel diesel and and gasoline gasoline is is more more 1. BACKGROUND 1. BACKGROUND suitable for PPCI combustion. Moreover, diesel and gasoline that fuel with characters between diesel and gasoline is more suitable for PPCI combustion. Moreover, diesel and gasoline that fuel with characters between diesel and gasoline is more 1. BACKGROUND 1. to BACKGROUND are two of the most widely used fuels around the world and suitable for PPCI combustion. Moreover, diesel and gasoline PPCI is considered be one of the most promising are two of the most widely used fuels around the world and suitable for PPCI combustion. Moreover, diesel and gasoline PPCI is considered to be one of the most promising can be easily purchased in commercial market. Nowadays, the most widely used fuels around the world and are two of combustion concepts. It has been widely studied by Reitz PPCI is considered to be one of the most promising can be easily purchased in commercial market. Nowadays, are two of the most widely used fuels around the world and combustion concepts. to It has widely by Reitz refineries around the world have been equipped with PPCI is considered be been one of the studied most promising (Hanson, R., Splitter, D. and Reitz, 2009, Ra, Y. et al, 2011), can be easily purchased in commercial market. Nowadays, combustion concepts. It has been widely studied by Reitz refineries around the world have been equipped with can be easily purchased in commercial market. Nowadays, (Hanson, R., Splitter, D. and Reitz, 2009, Ra, Y. et al, 2011), combustion concepts. It has been widely studied by Reitz complete refining facilities. Hence, “dieseline” could be a Yang al al, and around the have been with (Hanson, R., Splitter,F. D.et and Reitz, 2009, 2009, Ra, Y. Y.et etal al,(Sellnau 2011), refineries complete could be refineries refining around facilities. the world worldHence, have “dieseline” been equipped equipped witha Yang F F et etR., al (Yang, (Yang, F.D. et and al, 2013), 2013), and Sellnau Sellnau etet alal, (Sellnau (Hanson, Splitter, Reitz, Ra, 2011), good alternative before a more suitable fuel for PPCI et al, 2015) and has been proved more easily to realize and complete refining facilities. Hence, “dieseline” could be Yang F et al (Yang, F. et al, 2013), and Sellnau et al (Sellnau good alternative a Hence, more suitable fuel could for PPCI complete refining before facilities. “dieseline” be aa et al, F2015) hasF.been easily to and combustion Yang et al and (Yang, et al,proved 2013),more and Sellnau et realize al (Sellnau is supplied. In this diesel and control than HCCI. However, with further investigations, good before aa more fuel for et al, has proved more easily to is widely widely this study, study, and good alternative alternative beforesupplied. moreInsuitable suitable fuel diesel for PPCI PPCI control than and HCCI. However, et al, 2015) 2015) and has been been provedwith morefurther easily investigations, to realize realize and and combustion gasoline blended fuel is applied, whereas the control there are several difficulties in PPCI realization. Combustion combustion is widely supplied. In this study, diesel and control than HCCI. However, with further investigations, gasoline blended fuel is applied, whereas the control combustion is widely supplied. In this study, diesel and there arethan several difficulties in PPCI Combustion algorithm raised in this paper lays foundation for PPCI control HCCI. However, withrealization. further investigations, gasoline blended fuel is applied, whereas the control phase control is still the main obstacle of gasoline PPCI there are several difficulties in PPCI realization. Combustion algorithm raised in this paper lays foundation for PPCI gasoline blended fuel is applied, whereas the control phase are control is difficulties still the main obstacle of gasoline PPCI combustion control fueled with any fuel. there several in PPCI realization. Combustion algorithm this paper realization. Meanwhile, for diesel PPCI, hard for to phase control is gasoline PPCI controlin fuel.foundation algorithm raised raised infueled this with paperanylays lays foundation for for PPCI PPCI realization. Meanwhile, for main diesel obstacle PPCI, it’s it’sof for diesel diesel to combustion phase control is still still the the main obstacle ofhard gasoline PPCI combustion control fueled with any fuel. generate homogenous charge because of its poor vaporability. realization. Meanwhile, for diesel PPCI, it’s hard for diesel to combustion control fueled with any fuel. generate homogenous because of it’s its poor realization. Meanwhile,charge for diesel PPCI, hardvaporability. for diesel to FFE FFE (flexible (flexible fuel fuel engine) engine) is is regarded regarded as as aa frontier frontier and and hot hot Besides, the range of is limited. It generate poor Besides, the operating operatingcharge range because of PPCI PPCIof limited. It has has been been FFE generate homogenous homogenous charge because ofisits its poor vaporability. vaporability. topic of internal combustion engine research. Researches (flexible fuel engine) is regarded as a frontier and hot topic of internal combustion engine research. Researches FFE (flexible fuel engine) is regarded as a frontier and suggested that fuel with low octane number and good Besides, thethat operating rangelow of PPCI PPCI is limited. limited. Itand has good been show that some alternative fuels can not only optimize the hot suggestedthe fuel with octaneis number It Besides, operating range of has been topic of combustion Researches show some alternative fuelsengine can notresearch. only optimize the inintopic that of internal internal combustion engine research. Researches vaporability is ideal for PPCI combustion. Researches suggested that fuel with low octane number and good vaporabilitythat is fuel ideal with for low PPCIoctane combustion. suggested number Researches and good show cylinder combustion process of traditional gasoline and diesel some can optimize incylinder processfuels of traditional gasoline andthe diesel show that thatcombustion some alternative alternative fuels can not not only only optimize the inconducted by Manente et al (Manente, V. et al, 2009, 2010), vaporability is ideal for PPCI combustion. Researches conducted by is Manente V. et al, 2009, 2010), engine, vaporability ideal etforal (Manente, PPCI combustion. Researches but also the emission. FFE combined with cylinder combustion process of traditional gasoline and engine, also can can reduce reduce combined with cylinderbut combustion processthe of emission. traditionalFFE gasoline and diesel diesel Adhikary et al (Adhikary et al, 2012), Chang et al (Chang et conducted by Manente et al (Manente, V. et al, 2009, 2010), Adhikary (Adhikary et al, 2012), Chang et 2009, al (Chang et PPCI conductedetbyalManente et al (Manente, V. et al, 2010), provides a feasible solution to the but reduce the FFE with PPCI feasible to meeting meeting the ever ever strict strict engine,provides but also alsoacan can reducesolution the emission. emission. FFE combined combined with al, 2012), B. Wang et Wang et 2014), and Ciatti et engine, Adhikary (Adhikary et 2012), Chang et al, 2012), et B.al et al al (B. (B. Wang et al, al, 2014), and(Chang Ciatti et Adhikary et alWang (Adhikary et al, al, 2012), Chang et al al (Chang et emission regulation. PPCI provides a feasible solution to meeting the ever strict emission regulation. PPCI provides a feasible solution to meeting the ever strict al (Ciatti, S., and Subramanian, S. N., 2011) focused on PPCI al, 2012), B. Wang et et al, and Ciatti et al Subramanian, S. N., focused PPCI al,(Ciatti, 2012), S., B. and Wang et al al (B. (B. Wang Wang et 2011) al, 2014), 2014), and on Ciatti et emission regulation. emission regulation. combustion fueled with low octane number fuels. al (Ciatti, S., and Subramanian, S. N., 2011) focused on PPCI combustion with low octane number al (Ciatti, S.,fueled and Subramanian, S. N., 2011)fuels. focused on PPCI However, However, the the control control algorithm algorithm of of the the traditional traditional diesel diesel combustion fueled with low octane number fuels. engine is designed mainly based on open-loop control. The However, the control algorithm of the traditional combustion fueled with low octane et number fuels. et al, 2005) engine is designed mainly based on control.diesel The However, the control algorithm of open-loop the traditional diesel Research is conducted by Zhong al (Zhong Research is conducted by Zhong et al (Zhong et al, 2005) engine relationship between control parameters and performance is designed mainly based on open-loop control. The relationship between control parameters and performance engine is designed mainly based on open-loop control. The who focuses on designing fuel characters through blending Research is conducted conducted by Zhong Zhong et al al (Zhong (Zhong et al, al, 2005) indicators (such as power, economy, emission and so on) who focuses on designing fuel characters through blending Research is by et et 2005) relationship between control parameters and performance indicators as power, economy, emission so on) is is relationship(such between control parameters and and performance gasoline together with diesel, and the blended fuel has been who focuses on designing fuel characters through blending gasoline together with diesel, the blended fuel has been indicators who focuses on designing fueland characters through blending established qualitatively based on or experiment. as economy, emission so established qualitatively based on experience experience experiment. indicators (such (such as power, power, economy, emissionorand and so on) on) is is called “dieseline” firstly by Xu’s group (Rezaei, S. Z. et al, gasoline together with diesel, and the blended fuel has been called “dieseline” firstlydiesel, by Xu’s (Rezaei, S. has Z. etbeen al, Then, gasoline together with and group the blended fuel the control can and established based or Then, the qualitatively control parameters parameters can be be optimized optimized and established qualitatively based on on experience experience or experiment. experiment. 2011 and Zhang, F. et al, 2013). Adam Weall et al (Weall, A., called “dieseline” firstly by Xu’s group (Rezaei, S. Z. et al, 2011 Zhang, F.firstly et al, by 2013). et alS. (Weall, A., determined calledand “dieseline” Xu’sAdam groupWeall (Rezaei, Z. et al, by aa great quantity experimental calibration. Then, control be and determined greatparameters quantity of of can experimental calibration. Then, the the by control parameters can be optimized optimized and and Collings, N., 2007), Kalghatgi (Kalghatgi, G. T., 2011 and et Adam Weall A., and N.,F. Kalghatgi (Kalghatgi, G.(Weall, T., 2005) 2005) 2011Collings, and Zhang, Zhang, F.2007), et al, al, 2013). 2013). Adam Weall et et al al (Weall, A., Therefore, the alternative fuels (such as dieseline) could bring determined by a great quantity of experimental calibration. Therefore, the alternative fuels (such as dieseline) could bring determined by a great quantity of experimental calibration. and Shi et al (Shi, Y., and Reitz, R. D., 2010) conducted and 2007), Kalghatgi G. T., Shi et al N., (Shi, Y., and Reitz, (Kalghatgi, R. D., 2010) and Collings, Collings, N., 2007), Kalghatgi (Kalghatgi, G. conducted T., 2005) 2005) new problems to engine control system because Therefore, the fuels (such dieseline) could bring new problems to the the original original control system because Therefore, the alternative alternative fuelsengine (such as as dieseline) could bring research on “dieseline” as well and found it of good autoand Shi et al (Shi, Y., and Reitz, R. D., 2010) conducted research on al“dieseline” andR.found it of good auto- of and Shi et (Shi, Y., as andwell Reitz, D., 2010) conducted their different physical and chemical characters. new problems to the original engine control system because of their different physical and chemical characters. new problems to the original engine control system because ignition and evaporation property. It is generally accepted research on well it autoignition and evaporationas is generally accepted good auto- of their different physical and chemical characters. research on “dieseline” “dieseline” asproperty. well and andItfound found it of of good of their different physical and chemical characters. ignition ignition and and evaporation evaporation property. property. It It is is generally generally accepted accepted Copyright © 2016, 2016 IFAC 209Hosting by Elsevier Ltd. All rights reserved. 2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2016 IFAC 209 Peer review under responsibility of International Federation of Automatic Copyright © © 2016 2016 IFAC IFAC 209Control. Copyright 209 10.1016/j.ifacol.2016.08.031
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
203
generate different combustion indicators together for better accuracy. As shown in Fig. 2 (Wang, J., Yang, F. and Ouyang, M., 2015). The model is built based on statistical machine learning algorithm in ASCOMO. The inputs of the model are available from the control system. Of all the inputs, pedal and speed refer to operating condition, injection timing is one of the most important control parameters, IMEP and ignition delay are combustion indicators. Blend ratio is the output. Besides, power closed-loop control system and economy optimization method are proposed separately in this study.
There are two major problems for dieseline fueled FFE to overcome. The first problem is the output torque varies with dieseline of different diesel volumetric ratios. As shown in Fig. 1 (Wang, J., Yang, F., and Ouyang, M., 2015). The CFM (cycle fuel mass) is determined by MAP based on open-loop control. The output torque demand of the whole powertrain could be obtained through MAP according to the driver’s speed demand and pedal position. Then, the torque demand will be interpreted into CFM. However, the practical output torque will be different under the same load demand for dieseline of different diesel volumetric ratios because of their different lower heating values. As a result, the engine power performance cannot keep constant when using different fuels. Similarly, improper main injection timing will bring the possibility of efficiency deterioration or even misfire.
Firstly, this paper will discuss the necessity of fuel estimation model based feedforward controller to the control system. Next, a dieseline fueled flexible fuel engine accommodation closed-loop control algorithm is proposed aimed at optimizing power and economy performance simultaneously. Finally, this closed-loop control algorithm will be verified by simulation and the performance indicators of the system will be compared with that of the open-loop control algorithm. 2. EXPENRIMENTAL SETUP
Fig. 1. CFM open-loop control.
2.1 Fuel
Former researches which were conducted respectively by Snyder et al (Snyder et al, 2010), Zhao J et al (Zhao, J., and Wang, J., 2012), Mirheidari S et al (Mirheidari S et al, 2010), and Wang J et al (Wang, J., Neely, G. D., and Ryan, T. W., 2006) on diesel and bio-diesel blends fueled engine accommodation control tried to solve this problem mainly based on open-loop control algorithm through blend ratio estimation. The estimated blend ratio can be another dimension of MAP in engine electric control system. Thus, the control parameters are not only determined by driver‘s demand, but also by blend ratio. However, to precisely meet the demand of output torque and economy, the blend ratio estimation model should be very accurate, which could be difficult to realize and bring much more calibration effort.
SOC Indicator
Operating Condition
Control Parameter
Specifications of Beijing standard RON93 (research octane number) gasoline and 0# diesel are listed in Table 1 and Table 2 respectively. Table 1. Specifications of experimental gasoline Parameter RON Sulphur (m/m) Aromatics (V/V) 10% evaporation temperature (ºC) 50% evaporation temperature (ºC) 90% evaporation temperature (ºC) Density @20ºC (kg/m3) LHV (lower heating value) (J/ mm3)
Ignition Delay
IMEP Speed Pedal
Recognition Model of Blend Ratio Based on MultiParameters
Value 93 0.001% 35.5% 59.7 107.3 160.2 755.4 32.86
Table 2. Specifications of experimental diesel Blend Ratio
Parameter CN Sulphur (m/m) Aromatics (V/V) 10% evaporation temperature (ºC) 50% evaporation temperature (ºC) 90% evaporation temperature (ºC) Density @20ºC (kg/m3) LHV (lower heating value) (J/ mm3)
Main Injection
Fig. 2. Multi-indicator fusion blend ratio estimation model. The research group in Tsinghua University developed closedloop control algorithm, which was based on blend ratio online estimation. In the former study, the authors adapted multi-indicator fusion blend ratio estimation method to
Value 52.6 0.001% 29.6% 214.8 266.1 333.6 839.3 36.01
Blend ratio (BR) is defined as the volumetric ratio of diesel in the blended fuel, as shown in (1). The footnote “D” means diesel while “G” refers to gasoline. V is the volume of fuel. In 210
IFAC AAC 2016 204 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
terms of the definition, BR of pure gasoline is 0, while pure diesel is 1. To further investigate on this issue, we prepare fuels with BRs of 1, 0.5, 0.3 and 0.1, which can be named G0D100, G50D50, G70D30 and G90D10. BR VD / VG VD
angle and cycle. Closed-loop combustion control algorithm is also executed in iCAT. Feedbacks on control parameters are sent to the engine management core through a dedicated SPI (serial peripheral interface) bus from iCAT.
(1)
Table 5. Technical parameters of PSG
Table 3. Specifications of blended fuel Fuel G0D100 G50D50 G70D30 G90D10
BR 1 0.5 0.3 0.1
RON 27.0 60.0 73.2 86.4
Parameter Test range (MPa) Maximum Pressure (MPa) Temperature range (ºC) Power supply (V)
CN 52.6 33.1 25.3 17.6
In other words, the prototype ECU is composed of two cores, the iCAT core and the traditional engine management core. The iCAT core processes in-cylinder pressure sampling, combustion state indicators calculation and combustion feedback control tasks, while the engine management core conducts traditional engine management tasks.
Specifications of the blended fuels are listed in Table 3, where RON and CN of fuels are shown. The value of RON or CN can be calculated according to (2) (Morris, W., 2007).
CN=68.54-0.59 RON
Value 0-21 23 -40~140 3.3
(2)
2.2 Engine In this paper, experiments are conducted on a 4-cylinder 1.9L common rail diesel engine, which is equipped with VNT (variable nozzle turbo) and high pressure EGR (exhaust gas recirculation) system. The parameters of experimental engine are shown in Table 4 in detail. Table 4. Engine parameters Parameter Cylinder and structure Displacement (L) Bore (mm) Stroke (mm) Compression ratio Idle speed (rpm) Fuel injection system ECU Rated power (kW) Rated speed (rpm) Maximum Torque (N∙m)
Description Inline 4 1.905 81 92.4 17.5 800 Common rail EC4300D 93 4000 271
Fig. 3. Combustion closed-loop control system. 3. NECESSITY OF BR ESTIMATION MODEL BASED FEEDFORWARD CONTROLLER In our previous work, the necessity of feedforward controller which is based on BR estimation was not validated. If the previous control system operates well without BR estimation, feedforward controller and the corresponding calibration can be saved. However, considering the difference in lower heating value when using fuels of different BRs and the amount to compensate for, unacceptable power deterioration may appear under large fluctuation of BR. This section will discuss the role feedforward controller plays in power closedloop control system. To verify this issue, power closed-loop control system with and without feedforward controller are compared. The effect of feedforward controller on economy control system can be demonstrated in a similar way.
2.3 Combustion Closed-loop Control System Fig. 3 shows the closed-loop control system of the engine. The original ECU developed by Bosch is replaced by Tsinghua University self-developed prototype ECU EC4300D. Table 5 shows the technical parameters of PSG (pressure sensor glow-plug) sensor. It is a kind of massproduced resistance type pressure sensor, which is integrated with a glow-plug. An iCAT (in-cylinder combustion analysis tool) is integrated in the prototype ECU. The tool samples the in-cylinder pressure at every 0.2 degrees of crank angle. Combustion state indicators are calculated by iCAT by crank
The simulation result of the control system at variable load and constant load conditions is shown respectively in Fig. 4 and Fig. 5. The simulation is carried out at 1800 r/min. In variable load condition, the pedal position is changed linearly from 20% to 30% in 10 seconds while in constant load 211
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
condition, pedal is held at 20%. BR is changed from 1 to 0.1 at 5 seconds to test the response of the system. As shown in the legend of Fig. 5. Condition 1 refers to system in constant load condition with feedforward controller. Similarly, Condition 2 is constant load without feedforward controller. Condition 3 is variable load with feedforward controller. Condition 4 is variable load without feedforward controller.
Ttq i Vs m pmi /
load condition where the power deficiency will last through the whole process. From the sub-figure of Fig. 4, we can notice that it takes about extra 10s for CFM to transit from feedforward value to the exact target value in the system with feedforward controller. That is because BR estimation model can’t be absolutely accurate. In Fig. 5, IMEP of the system with feedforward controller can follow the target value well (shown as red dash line) both at variable and constant load. However, there still remains tiny IMEP deficiency because the estimated BR can’t be absolutely accurate. For system without feedforward controller, an obvious IMEP deficiency occurs. Besides, it takes a longer time to recover from IMEP deficiency, especially at variable load, which is unacceptable.
(3)
According to (3), where Ttq means output torque, i is the quantities of cylinders, Vs refers to the displacement of each cylinder, ηm indicates mechanical efficiency, τ is stroke, pmi refers to IMEP. For certain engine, Ttq is in proportion to IMEP. So IMEP can be chosen as the power indicator.
Feedforward controller will play a similar part in economy control system. Improper injection timing may bring cycle fluctuation or even misfire. In a word, control system without feedforward controller is sensitive to BR fluctuation, thus can’t meet the drivers’ demand. Therefore, feedforward controller based on BR estimation model is necessary to the closed-loop control system.
25 Cycle Fuel Mass (mg)
23 21
13.5
19
4. CLOSED-LOOP ACCOMMODATION CONTROL BASED ON FUEL ESTIMATION MODEL
13
17
12.5
15
5
10
15
20
25
In our previous work, power and economy closed-loop control algorithm were proposed and demonstrated separately. However, we can’t pursue a better power performance at the cost of economy deterioration. So it is necessary to put forward an integrated control algorithm which can balance power performance and economy simultaneously.
13 11
0
5
10 15 Time (s)
20
25
The control logic structure is shown in Fig. 6. The torque demand is interpreted into BMEP (brake mean effective pressure). FMEP (friction mean effective pressure) can be got through calibration. PMEP (pumping mean effective pressure) can be calculated by iCAT. Then, target IMEP can be obtained. The algorithm will be executed through closed-loop control of IMEP. Blend ratio is the output of feedforward controller, of which the accuracy has been validated in our previous work. In this section, we focus on how the system response to the error of feedforward controller instead of the controller itself. So BR is set artificially. CA50 (crank angle where 50% heat is released), which indicates combustion phase, is chosen as the closed-loop control target in economy control. At the point with the highest efficiency, CA50 is around 12º ATDC. Therefore, 12º ATDC can be set as the target CA50. However, it may not be exactly the highest efficiency when practical CA50 meets the target value. The algorithm can be optimized by amending target CA50 by introducing economy self-calibration controller, which is based on maximal-value-seeking method. The input of this controller is IMEP/CFM. According to (4), where ηi is indicated efficiency, gbm refers to CFM, Hu is the lower heating value. For certain engine and fuel, indicated efficiency is proportional to IMEP/CFM. In general, the algorithm is executed by adjusting the main injection timing to seek the most economy point on the premise that the engine power performance can be ensured.
Indicated Mean Effective Pressure (Bar)
Fig. 4. CFM of system with/without feedforward controller. 10 9 Condition 1 Condition 2 Condition 3 Condition 4
8 7 6 5 0
5
205
10 15 Time (s)
20
25
Fig. 5. IMEP of system with/without feedforward controller. As shown in Table 1 and Table 2, the lower heating value of diesel is higher than that of gasoline. The higher the BR is, the less CFM should be interpreted to meet the same torque demand. As shown in Fig. 4, CFM of the system with feedforward controller reaches the feedforward value instantly both at variable load and constant load. While in the system without feedforward controller, CFM rises to the target value slowly from the current value which leads to obvious IMEP deficiency. It is even more serious in variable
i pmi Vs / g bm H u 212
(4)
IFAC AAC 2016 206 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
As shown in Fig. 7. The system is demonstrated at 1800r/min, with pedal position changed linearly from 20% to 30% in 10 seconds and then back to 20% linearly in 5 seconds. The result is shown in the figures below. “OL” means open-loop and “CL” means closed-loop control. Considering the error of estimated BR, “negative” and “positive” means the BR is
negatively or positively estimated which means the estimated value is lower or higher than the real value. The simulation result is compared with that of the open-loop control algorithm so as to verify the consistency of the control system under the estimated BR error.
Fig. 6. Power performance and economy fusion closed-loop control system structure. 35
Pedal (%)
30 25 20 15
0
5
10
15
20
25 Time (s)
Fig. 7. Variation of pedal position. 213
30
35
40
45
50
June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
207
Cycle Oil Mass (mg)
24 CL Negative COM
22
OL Negative COM
13.8
20
13.6
18
13.4
16
0
5
14 12
0
5
10
15
20
25 Time (s)
30
35
40
45
50
Fig. 8. CFM of open-loop and closed-loop control system when BR is negatively estimated.
Cycle Oil Mass (mg)
24 CL Positive COM
22
OL Positive COM
20
13.6 13.55 13.5 13.45
18 16 14 12
0
0
5
13.7 13.6 13.5 13.4 40
5
10
15
20
25 Time (s)
30
35
40
41
42
45
50
45
50
Indicated Mean Effective Pressure (Bar)
Fig. 9. CFM of open-loop and closed-loop control system when BR is positively estimated. 10.5 10
CL Negtive IM EP CL Positive IM EP OL Negtive IM EP OL Positive IM EP Target IM EP
9 8 7 6 5 4.5
0
5
10
15
20
25 Time (s)
30
35
Fig. 10. IMEP of power performance and economy open/closed-loop control system.
214
40
43
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden 208
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
0.465 CL Negtive CL Positive OL Negtive OL Positive
Indicated Efficiency
0.46 0.455 0.45 0.445 0.44 0.435 0.43
0
5
10
15
20
25 Time (s)
30
35
40
45
50
Fig. 11. Indicated efficiency of power performance and economy open/closed-loop control system. Fig. 8 and Fig. 9 show the CFM of open-loop and closedloop control system when BR is negatively or positively estimated. The initial CFM curve shown in the subfigure is mainly influenced by two factors. On one hand, because of the error of estimated BR, additional fuel is introduced to compensate for the difference between initial CFM and target value. On the other hand, when the output torque reaches the target value, IMEP varies in a small range. During economy control process, indicated efficiency is optimized. So less CFM is needed to maintain the current IMEP. The influence of the first factor on CFM depends on the relationship of initial IMEP and target IMEP. In Fig. 8, BR is estimated negatively, so overmuch fuel is injected into the cylinder which leads to higher IMEP (shown in Fig. 10). In closedloop control system, to meet the target IMEP, practical CFM decreases rapidly because of the amendment to CFM. In Fig. 9, BR is estimated positively, so deficient fuel is supplied which leads to lower IMEP (shown in Fig. 10). But in closedloop control system, to meet the target IMEP, practical CFM rises rapidly because of the compensation for CFM. The second factor will reduce the CFM because of the optimization of indicated efficiency. At the initial time, the first factor is the main factor which has greater impact on practical CFM. Later, the second factor plays a more important role.
Fig. 11 shows the indicated efficiency of the engine. The simulation result shows that whether BR is estimated positively or negatively, the steady state indicated efficiency of closed-loop control system remains the same, which shows good consistency of the algorithm. For open-loop control system, the efficiency is badly affected by the error of fuel estimation model. Besides, the closed-loop system which adapts economy self-calibration controller has a higher efficiency than open-loop control system. As shown in Fig. 11, there is a slight decline of indicated efficiency at around 40s of closed-loop control system, which leads to the phenomenon shown in the right subfigure of Fig. 9. The practical CFM takes about extra 2 seconds to reach the final value. This is because the economy self-calibration algorithm is trying to explore the most economical point by comparing indicated efficiencies under different main injection timings. In conclusion, fuel estimation model based power and economy fusion closed-loop control system can optimize the indicated efficiency of engine. 5. CONCLUSION In this paper, the necessity of fuel estimation model based feedforward controller to closed-loop control algorithm is discussed. This issue is verified by comparing the response of the control system with feedforward controller to BR fluctuation with the system without the feedforward controller. Power and economy fusion accommodation closed-loop control algorithm is proposed to balance power performance and economy at the same time. The algorithm is demonstrated by simulation to verify its advantage over open -loop system on consistency and performance. The results are concluded as follows:
The practical IMEP and its target value are shown in Fig. 10. Whether the BR is estimated positively or negatively, IMEP of the closed-loop system follows the target value well. For open-loop control system, when BR is positively estimated, the practical IMEP is lower than the target value. When BR is negatively estimated, the practical IMEP is higher than the target value. The result indicates that the open-loop control system is more sensitive to fuel estimation model error than closed-loop control system. The closed-loop control system is proved to be of good consistency. In a word, fuel estimation model based closed-loop power and economy fusion accommodation control system can ensure the power performance of the engine.
① A power closed-loop control system based on IMEP without feedforward controller is established. Compared with the system with feedforward controller, results show that it takes a longer time for the system to recover from power deterioration, especially in variable load condition. The fuel estimation model based feedforward controller can speed up 215
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
system response characteristics, which proves to be necessary to closed-loop control system. ② Power and economy fusion accommodation closed-loop control algorithm with feedforward controller is established, aimed at balancing the power and economy performance of the flexible fuel engine at the same time. The results show that the proposed control system can ensure the power performance and the economy can also be optimized. Besides, compared with open-loop control system, closed-loop control system is less sensitive to the error of fuel estimation model, which shows better consistency. REFERENCES Adhikary, B. D., Ra, Y., Reitz, R. D., & Ciatti, S. (2012). Numerical Optimization of a Light-Duty Compression Ignition Engine Fuelled With Low-Octane Gasoline (No. 2012-01-1336). SAE Technical Paper. Chang, J., Kalghatgi, G., Amer, A., & Viollet, Y. (2012). Enabling high efficiency direct injection engine with naphtha fuel through Partially Premixed Charge Compression Ignition Combustion. SAE Technical paper, 01-0677. Ciatti, S., & Subramanian, S. N. (2011). An experimental investigation of low-octane gasoline in diesel engines. Journal of Engineering for Gas Turbines and Power, 133(9), 092802. Hanson, R., Splitter, D., & Reitz, R. D. (2009). Operating a heavy-duty direct-injection compression-ignition engine with gasoline for low emissions (No. 2009-01-1442). SAE Technical Paper. Kalghatgi, G. T. (2005). Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines. SAE paper, (2005-01), 0239. Kalghatgi, G. T., Risberg, P., & Ångström, H. E. (2007). Partially pre-mixed auto-ignition of gasoline to attain low smoke and low NOx at high load in a compression ignition engine and comparison with a diesel fuel. SAE paper, (2007-01), 0006. Manente, V., Johansson, B., & Tunestål, P. (2009). Partially premixed combustion at high load using gasoline and ethanol, a comparison with diesel. In SAE World Congress & Exhibition (Vol. 2009). Society of Automotive Engineers. Manente, V., Tunestal, P., Johansson, B., & Cannella, W. J. (2010). Effects of ethanol and different type of gasoline fuels on partially premixed combustion from low to high load (No. 2010-01-0871). SAE Technical Paper. Mirheidari, S., Mohammadpour, J., Grigoriadis, K. M., & Franchek, M. A. (2010, June). Biodiesel blend estimation based on fuel consumption and engine power. In American Control Conference (ACC), 2010 (pp. 30213026). IEEE. Morris, W. (2007). Method relates diesel cetane, octane ratings. Oil & gas journal, 105(45), 58-60. Ra, Y., Loeper, P., Reitz, R., Andrie, M., Krieger, R., Foster, D. ... & Szymkowicz, P. (2011). Study of high speed gasoline direct injection compression ignition (GDICI) engine operation in the LTC regime. SAE International Journal of Engines, 4(1), 1412-1430. 216
209
Rezaei, S. Z., Zhang, F., Xu, H., Ghafourian, A., Herreros, J. M., & Shuai, S. (2013). Investigation of two-stage splitinjection strategies for a Dieseline fuelled PPCI engine. Fuel, 107, 299-308. Sellnau, M., Moore, W., Sinnamon, J., Hoyer, K., Foster, M., & Husted, H. (2015). GDCI multi-cylinder engine for high fuel efficiency and low emissions. SAE International Journal of Engines, 8(2), 775-790. Shi, Y., & Reitz, R. D. (2010). Optimization of a heavy-duty compression–ignition engine fueled with diesel and gasoline-like fuels. Fuel, 89(11), 3416-3430. Snyder, D. B., Adi, G. H., Bunce, M. P., Satkoski, C. A., & Shaver, G. M. (2010). Fuel blend fraction estimation for fuel-flexible combustion control: Uncertainty analysis. Control Engineering Practice, 18(4), 418-432. Wang, B., Wang, Z., Shuai, S., Yang, H., & Wang, J. (2014). Combustion and emission characteristics of Multiple Premixed Compression Ignition (MPCI) fuelled with naphtha and gasoline in wide load range. Energy Conversion and Management, 88, 79-87. Wang, J., Neely, G. D., & Ryan, T. W. (2006). On-board fuel property classifier for fuel property adaptive engine control system (No. 2006-01-0054). SAE Technical Paper. Wang, J., Yang, F., & Ouyang, M. (2015). Dieseline fueled flexible fuel compression ignition engine control based on in-cylinder pressure sensor. Applied Energy, 159, 8796. Weall, A., & Collings, N. (2007). Investigation into partially premixed combustion in a light-duty multi-cylinder diesel engine fuelled gasoline and diesel with a mixture of. SAE Technical Papers. Yang, F., Gao, G., Ouyang, M., Chen, L., & Yang, Y. (2013). Research on a diesel HCCI engine assisted by an ISG motor. Applied Energy, 101, 718-729. Zhang, F., Xu, H., Zhang, J., Tian, G., & Kalghatgi, G. (2011). Investigation into light duty dieseline fuelled partially-premixed compression ignition engine. SAE International Journal of Engines, 4(1), 2124-2134. Zhao, J., & Wang, J. (2012, June). Energy-based and oxygenbased biodiesel blend level estimation methods for diesel engines. In American Control Conference (ACC), 2012 (pp. 4975-4980). IEEE. Zhong, S., Wyszynski, M. L., Megaritis, A., Yap, D., & Xu, H. (2005).Experimental investigation into HCCI combustion using gasoline and diesel blended fuels (No. 2005-01-3733). SAE Technical Paper.
ScienceDirect
IFAC-PapersOnLine 49-11 (2016) 202–209 Accommodation Accommodation Closed-loop Closed-loop Control Control of of Dieseline Dieseline Fueled Fueled Flexible Flexible Fuel Fuel Engine Engine Accommodation Control of Dieseline Fueled Flexible Fuel Engine Based on In-cylinder Pressure Sensor Accommodation Closed-loop Closed-loop Control of Dieseline Fueled Flexible Fuel Engine Based on In-cylinder Pressure Sensor Based on In-cylinder Pressure Sensor Based on In-cylinder Pressure Sensor Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Yang*.
Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Yang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. Yaodong Hu*. Tianyuan Zhou*. Changsheng Yao*. Fuyuan Minggao Haiyan Huang*. Yaodong Jinli Hu*.Wang*. Tianyuan Zhou*. Ouyang*. Changsheng Yao*. Fuyuan Yang*. Yang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. Jinli Wang*. Minggao Ouyang*. Haiyan Huang*. *State *State Key Key Laboratory Laboratory of of Automotive Automotive Safety Safety and and Energy, Energy, Tsinghua Tsinghua University, University, Beijing Beijing 100084, 100084, China China (e-mail: [email protected]). *State Key Laboratory of Automotive Safety and Energy, Tsinghua University, Beijing (e-mail: *State Key Laboratory of Automotive Safety [email protected]). and Energy, Tsinghua University, Beijing 100084, 100084, China China (e-mail: (e-mail: [email protected]). [email protected]).
Abstract: Abstract: This This paper paper conducts conducts investigation investigation on on accommodation accommodation closed-loop closed-loop control control of of dieseline dieseline fueled fueled flexible fuel engine. In our previous work, gasoline/diesel blend ratio estimation method was proposed proposed Abstract: This paper investigation on control of fueled flexible fuel engine. our previous work, gasoline/diesel blend closed-loop ratio estimation method was Abstract: This paperInconducts conducts investigation on accommodation accommodation closed-loop control of dieseline dieseline fueled and acted actedfuel as aaengine. feedforward controller. In this thisgasoline/diesel paper, power power performance performance closed-loopmethod controlwas systems with flexible In previous blend proposed and as feedforward In paper, closed-loop control systems with flexible fuel engine. In our ourcontroller. previous work, work, gasoline/diesel blend ratio ratio estimation estimation method was proposed or without blend ratio estimation model are simulated respectively at constant and variable load to andwithout acted as asblend feedforward controller. In this this paper, powerrespectively performanceatclosed-loop closed-loop control systems with or ratio estimation model arepaper, simulated constant and variable load to and acted aa feedforward controller. In power performance control systems with demonstrate the necessity of feedforward controller to control system. Results show that, whether at or without blend ratio estimation model are simulated respectively at constant and variable load to demonstrate the necessity of feedforward controller to control system. that, whether at or without blend ratio estimation model are simulated respectively at Results constantshow and variable load to constant or variable load, power indicators of the system with feedforward controller track the target demonstrate the necessity of feedforward controller to control system. Results show that, whether at constant or variable load, power indicatorscontroller of the system with system. feedforward controller trackwhether the target demonstrate the necessity of feedforward to control Results show that, at value well. well. However, forpower system without offeedforward feedforward controller, power controller performance deteriorates constant or load, indicators with track the value However, for system without power performance constant or variable variable load, power indicators of the the system system controller, with feedforward feedforward controller trackdeteriorates the target target obviously, especially especially in in variable loadwithout condition. Besides, accommodation accommodation closed-loop controldeteriorates system for for value for feedforward controller, performance obviously, load condition. Besides, closed-loop control system value well. well. However, However, variable for system system without feedforward controller, power power performance deteriorates dieseline fueled flexible fuel engine is established. The algorithm is executed by adjusting the main obviously, especially in variable load condition. Besides, accommodation closed-loop control system for dieseline flexible fuel engine is established. The accommodation algorithm is executed by adjusting the main obviously,fueled especially in variable load condition. Besides, closed-loop control system for injection timing timing to seek seek the most most economy economyestablished. point on on the theThe premise that the the power performance of the the engine dieseline fueled algorithm is executed by the main injection point premise that performance of dieseline fueled toflexible flexiblethefuel fuel engine engine is is established. The algorithm is power executed by adjusting adjusting theengine main can be be ensured ensured simultaneously. Compared with open-loop open-loop control algorithm, closed-loop control system injection timing to economy on that the performance of engine can with control injection timingsimultaneously. to seek seek the the most mostCompared economy point point on the the premise premise thatalgorithm, the power powerclosed-loop performancecontrol of the the system engine is less sensitive to fuel estimation model error, which proves to be of better consistency. Moreover, the canless be sensitive ensured simultaneously. simultaneously. Compared with which open-loop control algorithm, closed-loop control control system is to fuel estimation model error, proves to bealgorithm, of better consistency. Moreover, the can be ensured Compared with open-loop control closed-loop system practical output torque follows the target value well and the indicated efficiency can also be optimized. is less sensitive to fuel estimation model error, which proves to be of better consistency. Moreover, the practical output torque the target andproves the indicated can also beMoreover, optimized.the is less sensitive to fuelfollows estimation modelvalue error,well which to be ofefficiency better consistency. practical output torque follows the target value well and the indicated efficiency can also be optimized. practical output torque follows the target value well and the indicated efficiency can also be optimized. © 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: in-cylinder pressure sensor; closed-loop combustion control; dieseline; flexible fuel engine; Keywords: in-cylinder pressure sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation control;pressure economy optimization; PPCI; Keywords: in-cylinder sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation control;pressure economy optimization; PPCI; Keywords: in-cylinder sensor; closed-loop combustion control; dieseline; flexible fuel engine; accommodation accommodation control; control; economy economy optimization; optimization; PPCI; PPCI; that that fuel fuel with with characters characters between between diesel diesel and and gasoline gasoline is is more more 1. BACKGROUND 1. BACKGROUND suitable for PPCI combustion. Moreover, diesel and gasoline that fuel with characters between diesel and gasoline is more suitable for PPCI combustion. Moreover, diesel and gasoline that fuel with characters between diesel and gasoline is more 1. BACKGROUND 1. to BACKGROUND are two of the most widely used fuels around the world and suitable for PPCI combustion. Moreover, diesel and gasoline PPCI is considered be one of the most promising are two of the most widely used fuels around the world and suitable for PPCI combustion. Moreover, diesel and gasoline PPCI is considered to be one of the most promising can be easily purchased in commercial market. Nowadays, the most widely used fuels around the world and are two of combustion concepts. It has been widely studied by Reitz PPCI is considered to be one of the most promising can be easily purchased in commercial market. Nowadays, are two of the most widely used fuels around the world and combustion concepts. to It has widely by Reitz refineries around the world have been equipped with PPCI is considered be been one of the studied most promising (Hanson, R., Splitter, D. and Reitz, 2009, Ra, Y. et al, 2011), can be easily purchased in commercial market. Nowadays, combustion concepts. It has been widely studied by Reitz refineries around the world have been equipped with can be easily purchased in commercial market. Nowadays, (Hanson, R., Splitter, D. and Reitz, 2009, Ra, Y. et al, 2011), combustion concepts. It has been widely studied by Reitz complete refining facilities. Hence, “dieseline” could be a Yang al al, and around the have been with (Hanson, R., Splitter,F. D.et and Reitz, 2009, 2009, Ra, Y. Y.et etal al,(Sellnau 2011), refineries complete could be refineries refining around facilities. the world worldHence, have “dieseline” been equipped equipped witha Yang F F et etR., al (Yang, (Yang, F.D. et and al, 2013), 2013), and Sellnau Sellnau etet alal, (Sellnau (Hanson, Splitter, Reitz, Ra, 2011), good alternative before a more suitable fuel for PPCI et al, 2015) and has been proved more easily to realize and complete refining facilities. Hence, “dieseline” could be Yang F et al (Yang, F. et al, 2013), and Sellnau et al (Sellnau good alternative a Hence, more suitable fuel could for PPCI complete refining before facilities. “dieseline” be aa et al, F2015) hasF.been easily to and combustion Yang et al and (Yang, et al,proved 2013),more and Sellnau et realize al (Sellnau is supplied. In this diesel and control than HCCI. However, with further investigations, good before aa more fuel for et al, has proved more easily to is widely widely this study, study, and good alternative alternative beforesupplied. moreInsuitable suitable fuel diesel for PPCI PPCI control than and HCCI. However, et al, 2015) 2015) and has been been provedwith morefurther easily investigations, to realize realize and and combustion gasoline blended fuel is applied, whereas the control there are several difficulties in PPCI realization. Combustion combustion is widely supplied. In this study, diesel and control than HCCI. However, with further investigations, gasoline blended fuel is applied, whereas the control combustion is widely supplied. In this study, diesel and there arethan several difficulties in PPCI Combustion algorithm raised in this paper lays foundation for PPCI control HCCI. However, withrealization. further investigations, gasoline blended fuel is applied, whereas the control phase control is still the main obstacle of gasoline PPCI there are several difficulties in PPCI realization. Combustion algorithm raised in this paper lays foundation for PPCI gasoline blended fuel is applied, whereas the control phase are control is difficulties still the main obstacle of gasoline PPCI combustion control fueled with any fuel. there several in PPCI realization. Combustion algorithm this paper realization. Meanwhile, for diesel PPCI, hard for to phase control is gasoline PPCI controlin fuel.foundation algorithm raised raised infueled this with paperanylays lays foundation for for PPCI PPCI realization. Meanwhile, for main diesel obstacle PPCI, it’s it’sof for diesel diesel to combustion phase control is still still the the main obstacle ofhard gasoline PPCI combustion control fueled with any fuel. generate homogenous charge because of its poor vaporability. realization. Meanwhile, for diesel PPCI, it’s hard for diesel to combustion control fueled with any fuel. generate homogenous because of it’s its poor realization. Meanwhile,charge for diesel PPCI, hardvaporability. for diesel to FFE FFE (flexible (flexible fuel fuel engine) engine) is is regarded regarded as as aa frontier frontier and and hot hot Besides, the range of is limited. It generate poor Besides, the operating operatingcharge range because of PPCI PPCIof limited. It has has been been FFE generate homogenous homogenous charge because ofisits its poor vaporability. vaporability. topic of internal combustion engine research. Researches (flexible fuel engine) is regarded as a frontier and hot topic of internal combustion engine research. Researches FFE (flexible fuel engine) is regarded as a frontier and suggested that fuel with low octane number and good Besides, thethat operating rangelow of PPCI PPCI is limited. limited. Itand has good been show that some alternative fuels can not only optimize the hot suggestedthe fuel with octaneis number It Besides, operating range of has been topic of combustion Researches show some alternative fuelsengine can notresearch. only optimize the inintopic that of internal internal combustion engine research. Researches vaporability is ideal for PPCI combustion. Researches suggested that fuel with low octane number and good vaporabilitythat is fuel ideal with for low PPCIoctane combustion. suggested number Researches and good show cylinder combustion process of traditional gasoline and diesel some can optimize incylinder processfuels of traditional gasoline andthe diesel show that thatcombustion some alternative alternative fuels can not not only only optimize the inconducted by Manente et al (Manente, V. et al, 2009, 2010), vaporability is ideal for PPCI combustion. Researches conducted by is Manente V. et al, 2009, 2010), engine, vaporability ideal etforal (Manente, PPCI combustion. Researches but also the emission. FFE combined with cylinder combustion process of traditional gasoline and engine, also can can reduce reduce combined with cylinderbut combustion processthe of emission. traditionalFFE gasoline and diesel diesel Adhikary et al (Adhikary et al, 2012), Chang et al (Chang et conducted by Manente et al (Manente, V. et al, 2009, 2010), Adhikary (Adhikary et al, 2012), Chang et 2009, al (Chang et PPCI conductedetbyalManente et al (Manente, V. et al, 2010), provides a feasible solution to the but reduce the FFE with PPCI feasible to meeting meeting the ever ever strict strict engine,provides but also alsoacan can reducesolution the emission. emission. FFE combined combined with al, 2012), B. Wang et Wang et 2014), and Ciatti et engine, Adhikary (Adhikary et 2012), Chang et al, 2012), et B.al et al al (B. (B. Wang et al, al, 2014), and(Chang Ciatti et Adhikary et alWang (Adhikary et al, al, 2012), Chang et al al (Chang et emission regulation. PPCI provides a feasible solution to meeting the ever strict emission regulation. PPCI provides a feasible solution to meeting the ever strict al (Ciatti, S., and Subramanian, S. N., 2011) focused on PPCI al, 2012), B. Wang et et al, and Ciatti et al Subramanian, S. N., focused PPCI al,(Ciatti, 2012), S., B. and Wang et al al (B. (B. Wang Wang et 2011) al, 2014), 2014), and on Ciatti et emission regulation. emission regulation. combustion fueled with low octane number fuels. al (Ciatti, S., and Subramanian, S. N., 2011) focused on PPCI combustion with low octane number al (Ciatti, S.,fueled and Subramanian, S. N., 2011)fuels. focused on PPCI However, However, the the control control algorithm algorithm of of the the traditional traditional diesel diesel combustion fueled with low octane number fuels. engine is designed mainly based on open-loop control. The However, the control algorithm of the traditional combustion fueled with low octane et number fuels. et al, 2005) engine is designed mainly based on control.diesel The However, the control algorithm of open-loop the traditional diesel Research is conducted by Zhong al (Zhong Research is conducted by Zhong et al (Zhong et al, 2005) engine relationship between control parameters and performance is designed mainly based on open-loop control. The relationship between control parameters and performance engine is designed mainly based on open-loop control. The who focuses on designing fuel characters through blending Research is conducted conducted by Zhong Zhong et al al (Zhong (Zhong et al, al, 2005) indicators (such as power, economy, emission and so on) who focuses on designing fuel characters through blending Research is by et et 2005) relationship between control parameters and performance indicators as power, economy, emission so on) is is relationship(such between control parameters and and performance gasoline together with diesel, and the blended fuel has been who focuses on designing fuel characters through blending gasoline together with diesel, the blended fuel has been indicators who focuses on designing fueland characters through blending established qualitatively based on or experiment. as economy, emission so established qualitatively based on experience experience experiment. indicators (such (such as power, power, economy, emissionorand and so on) on) is is called “dieseline” firstly by Xu’s group (Rezaei, S. Z. et al, gasoline together with diesel, and the blended fuel has been called “dieseline” firstlydiesel, by Xu’s (Rezaei, S. has Z. etbeen al, Then, gasoline together with and group the blended fuel the control can and established based or Then, the qualitatively control parameters parameters can be be optimized optimized and established qualitatively based on on experience experience or experiment. experiment. 2011 and Zhang, F. et al, 2013). Adam Weall et al (Weall, A., called “dieseline” firstly by Xu’s group (Rezaei, S. Z. et al, 2011 Zhang, F.firstly et al, by 2013). et alS. (Weall, A., determined calledand “dieseline” Xu’sAdam groupWeall (Rezaei, Z. et al, by aa great quantity experimental calibration. Then, control be and determined greatparameters quantity of of can experimental calibration. Then, the the by control parameters can be optimized optimized and and Collings, N., 2007), Kalghatgi (Kalghatgi, G. T., 2011 and et Adam Weall A., and N.,F. Kalghatgi (Kalghatgi, G.(Weall, T., 2005) 2005) 2011Collings, and Zhang, Zhang, F.2007), et al, al, 2013). 2013). Adam Weall et et al al (Weall, A., Therefore, the alternative fuels (such as dieseline) could bring determined by a great quantity of experimental calibration. Therefore, the alternative fuels (such as dieseline) could bring determined by a great quantity of experimental calibration. and Shi et al (Shi, Y., and Reitz, R. D., 2010) conducted and 2007), Kalghatgi G. T., Shi et al N., (Shi, Y., and Reitz, (Kalghatgi, R. D., 2010) and Collings, Collings, N., 2007), Kalghatgi (Kalghatgi, G. conducted T., 2005) 2005) new problems to engine control system because Therefore, the fuels (such dieseline) could bring new problems to the the original original control system because Therefore, the alternative alternative fuelsengine (such as as dieseline) could bring research on “dieseline” as well and found it of good autoand Shi et al (Shi, Y., and Reitz, R. D., 2010) conducted research on al“dieseline” andR.found it of good auto- of and Shi et (Shi, Y., as andwell Reitz, D., 2010) conducted their different physical and chemical characters. new problems to the original engine control system because of their different physical and chemical characters. new problems to the original engine control system because ignition and evaporation property. It is generally accepted research on well it autoignition and evaporationas is generally accepted good auto- of their different physical and chemical characters. research on “dieseline” “dieseline” asproperty. well and andItfound found it of of good of their different physical and chemical characters. ignition ignition and and evaporation evaporation property. property. It It is is generally generally accepted accepted Copyright © 2016, 2016 IFAC 209Hosting by Elsevier Ltd. All rights reserved. 2405-8963 © IFAC (International Federation of Automatic Control) Copyright © 2016 IFAC 209 Peer review under responsibility of International Federation of Automatic Copyright © © 2016 2016 IFAC IFAC 209Control. Copyright 209 10.1016/j.ifacol.2016.08.031
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
203
generate different combustion indicators together for better accuracy. As shown in Fig. 2 (Wang, J., Yang, F. and Ouyang, M., 2015). The model is built based on statistical machine learning algorithm in ASCOMO. The inputs of the model are available from the control system. Of all the inputs, pedal and speed refer to operating condition, injection timing is one of the most important control parameters, IMEP and ignition delay are combustion indicators. Blend ratio is the output. Besides, power closed-loop control system and economy optimization method are proposed separately in this study.
There are two major problems for dieseline fueled FFE to overcome. The first problem is the output torque varies with dieseline of different diesel volumetric ratios. As shown in Fig. 1 (Wang, J., Yang, F., and Ouyang, M., 2015). The CFM (cycle fuel mass) is determined by MAP based on open-loop control. The output torque demand of the whole powertrain could be obtained through MAP according to the driver’s speed demand and pedal position. Then, the torque demand will be interpreted into CFM. However, the practical output torque will be different under the same load demand for dieseline of different diesel volumetric ratios because of their different lower heating values. As a result, the engine power performance cannot keep constant when using different fuels. Similarly, improper main injection timing will bring the possibility of efficiency deterioration or even misfire.
Firstly, this paper will discuss the necessity of fuel estimation model based feedforward controller to the control system. Next, a dieseline fueled flexible fuel engine accommodation closed-loop control algorithm is proposed aimed at optimizing power and economy performance simultaneously. Finally, this closed-loop control algorithm will be verified by simulation and the performance indicators of the system will be compared with that of the open-loop control algorithm. 2. EXPENRIMENTAL SETUP
Fig. 1. CFM open-loop control.
2.1 Fuel
Former researches which were conducted respectively by Snyder et al (Snyder et al, 2010), Zhao J et al (Zhao, J., and Wang, J., 2012), Mirheidari S et al (Mirheidari S et al, 2010), and Wang J et al (Wang, J., Neely, G. D., and Ryan, T. W., 2006) on diesel and bio-diesel blends fueled engine accommodation control tried to solve this problem mainly based on open-loop control algorithm through blend ratio estimation. The estimated blend ratio can be another dimension of MAP in engine electric control system. Thus, the control parameters are not only determined by driver‘s demand, but also by blend ratio. However, to precisely meet the demand of output torque and economy, the blend ratio estimation model should be very accurate, which could be difficult to realize and bring much more calibration effort.
SOC Indicator
Operating Condition
Control Parameter
Specifications of Beijing standard RON93 (research octane number) gasoline and 0# diesel are listed in Table 1 and Table 2 respectively. Table 1. Specifications of experimental gasoline Parameter RON Sulphur (m/m) Aromatics (V/V) 10% evaporation temperature (ºC) 50% evaporation temperature (ºC) 90% evaporation temperature (ºC) Density @20ºC (kg/m3) LHV (lower heating value) (J/ mm3)
Ignition Delay
IMEP Speed Pedal
Recognition Model of Blend Ratio Based on MultiParameters
Value 93 0.001% 35.5% 59.7 107.3 160.2 755.4 32.86
Table 2. Specifications of experimental diesel Blend Ratio
Parameter CN Sulphur (m/m) Aromatics (V/V) 10% evaporation temperature (ºC) 50% evaporation temperature (ºC) 90% evaporation temperature (ºC) Density @20ºC (kg/m3) LHV (lower heating value) (J/ mm3)
Main Injection
Fig. 2. Multi-indicator fusion blend ratio estimation model. The research group in Tsinghua University developed closedloop control algorithm, which was based on blend ratio online estimation. In the former study, the authors adapted multi-indicator fusion blend ratio estimation method to
Value 52.6 0.001% 29.6% 214.8 266.1 333.6 839.3 36.01
Blend ratio (BR) is defined as the volumetric ratio of diesel in the blended fuel, as shown in (1). The footnote “D” means diesel while “G” refers to gasoline. V is the volume of fuel. In 210
IFAC AAC 2016 204 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
terms of the definition, BR of pure gasoline is 0, while pure diesel is 1. To further investigate on this issue, we prepare fuels with BRs of 1, 0.5, 0.3 and 0.1, which can be named G0D100, G50D50, G70D30 and G90D10. BR VD / VG VD
angle and cycle. Closed-loop combustion control algorithm is also executed in iCAT. Feedbacks on control parameters are sent to the engine management core through a dedicated SPI (serial peripheral interface) bus from iCAT.
(1)
Table 5. Technical parameters of PSG
Table 3. Specifications of blended fuel Fuel G0D100 G50D50 G70D30 G90D10
BR 1 0.5 0.3 0.1
RON 27.0 60.0 73.2 86.4
Parameter Test range (MPa) Maximum Pressure (MPa) Temperature range (ºC) Power supply (V)
CN 52.6 33.1 25.3 17.6
In other words, the prototype ECU is composed of two cores, the iCAT core and the traditional engine management core. The iCAT core processes in-cylinder pressure sampling, combustion state indicators calculation and combustion feedback control tasks, while the engine management core conducts traditional engine management tasks.
Specifications of the blended fuels are listed in Table 3, where RON and CN of fuels are shown. The value of RON or CN can be calculated according to (2) (Morris, W., 2007).
CN=68.54-0.59 RON
Value 0-21 23 -40~140 3.3
(2)
2.2 Engine In this paper, experiments are conducted on a 4-cylinder 1.9L common rail diesel engine, which is equipped with VNT (variable nozzle turbo) and high pressure EGR (exhaust gas recirculation) system. The parameters of experimental engine are shown in Table 4 in detail. Table 4. Engine parameters Parameter Cylinder and structure Displacement (L) Bore (mm) Stroke (mm) Compression ratio Idle speed (rpm) Fuel injection system ECU Rated power (kW) Rated speed (rpm) Maximum Torque (N∙m)
Description Inline 4 1.905 81 92.4 17.5 800 Common rail EC4300D 93 4000 271
Fig. 3. Combustion closed-loop control system. 3. NECESSITY OF BR ESTIMATION MODEL BASED FEEDFORWARD CONTROLLER In our previous work, the necessity of feedforward controller which is based on BR estimation was not validated. If the previous control system operates well without BR estimation, feedforward controller and the corresponding calibration can be saved. However, considering the difference in lower heating value when using fuels of different BRs and the amount to compensate for, unacceptable power deterioration may appear under large fluctuation of BR. This section will discuss the role feedforward controller plays in power closedloop control system. To verify this issue, power closed-loop control system with and without feedforward controller are compared. The effect of feedforward controller on economy control system can be demonstrated in a similar way.
2.3 Combustion Closed-loop Control System Fig. 3 shows the closed-loop control system of the engine. The original ECU developed by Bosch is replaced by Tsinghua University self-developed prototype ECU EC4300D. Table 5 shows the technical parameters of PSG (pressure sensor glow-plug) sensor. It is a kind of massproduced resistance type pressure sensor, which is integrated with a glow-plug. An iCAT (in-cylinder combustion analysis tool) is integrated in the prototype ECU. The tool samples the in-cylinder pressure at every 0.2 degrees of crank angle. Combustion state indicators are calculated by iCAT by crank
The simulation result of the control system at variable load and constant load conditions is shown respectively in Fig. 4 and Fig. 5. The simulation is carried out at 1800 r/min. In variable load condition, the pedal position is changed linearly from 20% to 30% in 10 seconds while in constant load 211
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
condition, pedal is held at 20%. BR is changed from 1 to 0.1 at 5 seconds to test the response of the system. As shown in the legend of Fig. 5. Condition 1 refers to system in constant load condition with feedforward controller. Similarly, Condition 2 is constant load without feedforward controller. Condition 3 is variable load with feedforward controller. Condition 4 is variable load without feedforward controller.
Ttq i Vs m pmi /
load condition where the power deficiency will last through the whole process. From the sub-figure of Fig. 4, we can notice that it takes about extra 10s for CFM to transit from feedforward value to the exact target value in the system with feedforward controller. That is because BR estimation model can’t be absolutely accurate. In Fig. 5, IMEP of the system with feedforward controller can follow the target value well (shown as red dash line) both at variable and constant load. However, there still remains tiny IMEP deficiency because the estimated BR can’t be absolutely accurate. For system without feedforward controller, an obvious IMEP deficiency occurs. Besides, it takes a longer time to recover from IMEP deficiency, especially at variable load, which is unacceptable.
(3)
According to (3), where Ttq means output torque, i is the quantities of cylinders, Vs refers to the displacement of each cylinder, ηm indicates mechanical efficiency, τ is stroke, pmi refers to IMEP. For certain engine, Ttq is in proportion to IMEP. So IMEP can be chosen as the power indicator.
Feedforward controller will play a similar part in economy control system. Improper injection timing may bring cycle fluctuation or even misfire. In a word, control system without feedforward controller is sensitive to BR fluctuation, thus can’t meet the drivers’ demand. Therefore, feedforward controller based on BR estimation model is necessary to the closed-loop control system.
25 Cycle Fuel Mass (mg)
23 21
13.5
19
4. CLOSED-LOOP ACCOMMODATION CONTROL BASED ON FUEL ESTIMATION MODEL
13
17
12.5
15
5
10
15
20
25
In our previous work, power and economy closed-loop control algorithm were proposed and demonstrated separately. However, we can’t pursue a better power performance at the cost of economy deterioration. So it is necessary to put forward an integrated control algorithm which can balance power performance and economy simultaneously.
13 11
0
5
10 15 Time (s)
20
25
The control logic structure is shown in Fig. 6. The torque demand is interpreted into BMEP (brake mean effective pressure). FMEP (friction mean effective pressure) can be got through calibration. PMEP (pumping mean effective pressure) can be calculated by iCAT. Then, target IMEP can be obtained. The algorithm will be executed through closed-loop control of IMEP. Blend ratio is the output of feedforward controller, of which the accuracy has been validated in our previous work. In this section, we focus on how the system response to the error of feedforward controller instead of the controller itself. So BR is set artificially. CA50 (crank angle where 50% heat is released), which indicates combustion phase, is chosen as the closed-loop control target in economy control. At the point with the highest efficiency, CA50 is around 12º ATDC. Therefore, 12º ATDC can be set as the target CA50. However, it may not be exactly the highest efficiency when practical CA50 meets the target value. The algorithm can be optimized by amending target CA50 by introducing economy self-calibration controller, which is based on maximal-value-seeking method. The input of this controller is IMEP/CFM. According to (4), where ηi is indicated efficiency, gbm refers to CFM, Hu is the lower heating value. For certain engine and fuel, indicated efficiency is proportional to IMEP/CFM. In general, the algorithm is executed by adjusting the main injection timing to seek the most economy point on the premise that the engine power performance can be ensured.
Indicated Mean Effective Pressure (Bar)
Fig. 4. CFM of system with/without feedforward controller. 10 9 Condition 1 Condition 2 Condition 3 Condition 4
8 7 6 5 0
5
205
10 15 Time (s)
20
25
Fig. 5. IMEP of system with/without feedforward controller. As shown in Table 1 and Table 2, the lower heating value of diesel is higher than that of gasoline. The higher the BR is, the less CFM should be interpreted to meet the same torque demand. As shown in Fig. 4, CFM of the system with feedforward controller reaches the feedforward value instantly both at variable load and constant load. While in the system without feedforward controller, CFM rises to the target value slowly from the current value which leads to obvious IMEP deficiency. It is even more serious in variable
i pmi Vs / g bm H u 212
(4)
IFAC AAC 2016 206 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
As shown in Fig. 7. The system is demonstrated at 1800r/min, with pedal position changed linearly from 20% to 30% in 10 seconds and then back to 20% linearly in 5 seconds. The result is shown in the figures below. “OL” means open-loop and “CL” means closed-loop control. Considering the error of estimated BR, “negative” and “positive” means the BR is
negatively or positively estimated which means the estimated value is lower or higher than the real value. The simulation result is compared with that of the open-loop control algorithm so as to verify the consistency of the control system under the estimated BR error.
Fig. 6. Power performance and economy fusion closed-loop control system structure. 35
Pedal (%)
30 25 20 15
0
5
10
15
20
25 Time (s)
Fig. 7. Variation of pedal position. 213
30
35
40
45
50
June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
207
Cycle Oil Mass (mg)
24 CL Negative COM
22
OL Negative COM
13.8
20
13.6
18
13.4
16
0
5
14 12
0
5
10
15
20
25 Time (s)
30
35
40
45
50
Fig. 8. CFM of open-loop and closed-loop control system when BR is negatively estimated.
Cycle Oil Mass (mg)
24 CL Positive COM
22
OL Positive COM
20
13.6 13.55 13.5 13.45
18 16 14 12
0
0
5
13.7 13.6 13.5 13.4 40
5
10
15
20
25 Time (s)
30
35
40
41
42
45
50
45
50
Indicated Mean Effective Pressure (Bar)
Fig. 9. CFM of open-loop and closed-loop control system when BR is positively estimated. 10.5 10
CL Negtive IM EP CL Positive IM EP OL Negtive IM EP OL Positive IM EP Target IM EP
9 8 7 6 5 4.5
0
5
10
15
20
25 Time (s)
30
35
Fig. 10. IMEP of power performance and economy open/closed-loop control system.
214
40
43
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden 208
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
0.465 CL Negtive CL Positive OL Negtive OL Positive
Indicated Efficiency
0.46 0.455 0.45 0.445 0.44 0.435 0.43
0
5
10
15
20
25 Time (s)
30
35
40
45
50
Fig. 11. Indicated efficiency of power performance and economy open/closed-loop control system. Fig. 8 and Fig. 9 show the CFM of open-loop and closedloop control system when BR is negatively or positively estimated. The initial CFM curve shown in the subfigure is mainly influenced by two factors. On one hand, because of the error of estimated BR, additional fuel is introduced to compensate for the difference between initial CFM and target value. On the other hand, when the output torque reaches the target value, IMEP varies in a small range. During economy control process, indicated efficiency is optimized. So less CFM is needed to maintain the current IMEP. The influence of the first factor on CFM depends on the relationship of initial IMEP and target IMEP. In Fig. 8, BR is estimated negatively, so overmuch fuel is injected into the cylinder which leads to higher IMEP (shown in Fig. 10). In closedloop control system, to meet the target IMEP, practical CFM decreases rapidly because of the amendment to CFM. In Fig. 9, BR is estimated positively, so deficient fuel is supplied which leads to lower IMEP (shown in Fig. 10). But in closedloop control system, to meet the target IMEP, practical CFM rises rapidly because of the compensation for CFM. The second factor will reduce the CFM because of the optimization of indicated efficiency. At the initial time, the first factor is the main factor which has greater impact on practical CFM. Later, the second factor plays a more important role.
Fig. 11 shows the indicated efficiency of the engine. The simulation result shows that whether BR is estimated positively or negatively, the steady state indicated efficiency of closed-loop control system remains the same, which shows good consistency of the algorithm. For open-loop control system, the efficiency is badly affected by the error of fuel estimation model. Besides, the closed-loop system which adapts economy self-calibration controller has a higher efficiency than open-loop control system. As shown in Fig. 11, there is a slight decline of indicated efficiency at around 40s of closed-loop control system, which leads to the phenomenon shown in the right subfigure of Fig. 9. The practical CFM takes about extra 2 seconds to reach the final value. This is because the economy self-calibration algorithm is trying to explore the most economical point by comparing indicated efficiencies under different main injection timings. In conclusion, fuel estimation model based power and economy fusion closed-loop control system can optimize the indicated efficiency of engine. 5. CONCLUSION In this paper, the necessity of fuel estimation model based feedforward controller to closed-loop control algorithm is discussed. This issue is verified by comparing the response of the control system with feedforward controller to BR fluctuation with the system without the feedforward controller. Power and economy fusion accommodation closed-loop control algorithm is proposed to balance power performance and economy at the same time. The algorithm is demonstrated by simulation to verify its advantage over open -loop system on consistency and performance. The results are concluded as follows:
The practical IMEP and its target value are shown in Fig. 10. Whether the BR is estimated positively or negatively, IMEP of the closed-loop system follows the target value well. For open-loop control system, when BR is positively estimated, the practical IMEP is lower than the target value. When BR is negatively estimated, the practical IMEP is higher than the target value. The result indicates that the open-loop control system is more sensitive to fuel estimation model error than closed-loop control system. The closed-loop control system is proved to be of good consistency. In a word, fuel estimation model based closed-loop power and economy fusion accommodation control system can ensure the power performance of the engine.
① A power closed-loop control system based on IMEP without feedforward controller is established. Compared with the system with feedforward controller, results show that it takes a longer time for the system to recover from power deterioration, especially in variable load condition. The fuel estimation model based feedforward controller can speed up 215
IFAC AAC 2016 June 19-23, 2016. Norrköping, Sweden
Yaodong Hu et al. / IFAC-PapersOnLine 49-11 (2016) 202–209
system response characteristics, which proves to be necessary to closed-loop control system. ② Power and economy fusion accommodation closed-loop control algorithm with feedforward controller is established, aimed at balancing the power and economy performance of the flexible fuel engine at the same time. The results show that the proposed control system can ensure the power performance and the economy can also be optimized. Besides, compared with open-loop control system, closed-loop control system is less sensitive to the error of fuel estimation model, which shows better consistency. REFERENCES Adhikary, B. D., Ra, Y., Reitz, R. D., & Ciatti, S. (2012). Numerical Optimization of a Light-Duty Compression Ignition Engine Fuelled With Low-Octane Gasoline (No. 2012-01-1336). SAE Technical Paper. Chang, J., Kalghatgi, G., Amer, A., & Viollet, Y. (2012). Enabling high efficiency direct injection engine with naphtha fuel through Partially Premixed Charge Compression Ignition Combustion. SAE Technical paper, 01-0677. Ciatti, S., & Subramanian, S. N. (2011). An experimental investigation of low-octane gasoline in diesel engines. Journal of Engineering for Gas Turbines and Power, 133(9), 092802. Hanson, R., Splitter, D., & Reitz, R. D. (2009). Operating a heavy-duty direct-injection compression-ignition engine with gasoline for low emissions (No. 2009-01-1442). SAE Technical Paper. Kalghatgi, G. T. (2005). Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines. SAE paper, (2005-01), 0239. Kalghatgi, G. T., Risberg, P., & Ångström, H. E. (2007). Partially pre-mixed auto-ignition of gasoline to attain low smoke and low NOx at high load in a compression ignition engine and comparison with a diesel fuel. SAE paper, (2007-01), 0006. Manente, V., Johansson, B., & Tunestål, P. (2009). Partially premixed combustion at high load using gasoline and ethanol, a comparison with diesel. In SAE World Congress & Exhibition (Vol. 2009). Society of Automotive Engineers. Manente, V., Tunestal, P., Johansson, B., & Cannella, W. J. (2010). Effects of ethanol and different type of gasoline fuels on partially premixed combustion from low to high load (No. 2010-01-0871). SAE Technical Paper. Mirheidari, S., Mohammadpour, J., Grigoriadis, K. M., & Franchek, M. A. (2010, June). Biodiesel blend estimation based on fuel consumption and engine power. In American Control Conference (ACC), 2010 (pp. 30213026). IEEE. Morris, W. (2007). Method relates diesel cetane, octane ratings. Oil & gas journal, 105(45), 58-60. Ra, Y., Loeper, P., Reitz, R., Andrie, M., Krieger, R., Foster, D. ... & Szymkowicz, P. (2011). Study of high speed gasoline direct injection compression ignition (GDICI) engine operation in the LTC regime. SAE International Journal of Engines, 4(1), 1412-1430. 216
209
Rezaei, S. Z., Zhang, F., Xu, H., Ghafourian, A., Herreros, J. M., & Shuai, S. (2013). Investigation of two-stage splitinjection strategies for a Dieseline fuelled PPCI engine. Fuel, 107, 299-308. Sellnau, M., Moore, W., Sinnamon, J., Hoyer, K., Foster, M., & Husted, H. (2015). GDCI multi-cylinder engine for high fuel efficiency and low emissions. SAE International Journal of Engines, 8(2), 775-790. Shi, Y., & Reitz, R. D. (2010). Optimization of a heavy-duty compression–ignition engine fueled with diesel and gasoline-like fuels. Fuel, 89(11), 3416-3430. Snyder, D. B., Adi, G. H., Bunce, M. P., Satkoski, C. A., & Shaver, G. M. (2010). Fuel blend fraction estimation for fuel-flexible combustion control: Uncertainty analysis. Control Engineering Practice, 18(4), 418-432. Wang, B., Wang, Z., Shuai, S., Yang, H., & Wang, J. (2014). Combustion and emission characteristics of Multiple Premixed Compression Ignition (MPCI) fuelled with naphtha and gasoline in wide load range. Energy Conversion and Management, 88, 79-87. Wang, J., Neely, G. D., & Ryan, T. W. (2006). On-board fuel property classifier for fuel property adaptive engine control system (No. 2006-01-0054). SAE Technical Paper. Wang, J., Yang, F., & Ouyang, M. (2015). Dieseline fueled flexible fuel compression ignition engine control based on in-cylinder pressure sensor. Applied Energy, 159, 8796. Weall, A., & Collings, N. (2007). Investigation into partially premixed combustion in a light-duty multi-cylinder diesel engine fuelled gasoline and diesel with a mixture of. SAE Technical Papers. Yang, F., Gao, G., Ouyang, M., Chen, L., & Yang, Y. (2013). Research on a diesel HCCI engine assisted by an ISG motor. Applied Energy, 101, 718-729. Zhang, F., Xu, H., Zhang, J., Tian, G., & Kalghatgi, G. (2011). Investigation into light duty dieseline fuelled partially-premixed compression ignition engine. SAE International Journal of Engines, 4(1), 2124-2134. Zhao, J., & Wang, J. (2012, June). Energy-based and oxygenbased biodiesel blend level estimation methods for diesel engines. In American Control Conference (ACC), 2012 (pp. 4975-4980). IEEE. Zhong, S., Wyszynski, M. L., Megaritis, A., Yap, D., & Xu, H. (2005).Experimental investigation into HCCI combustion using gasoline and diesel blended fuels (No. 2005-01-3733). SAE Technical Paper.