DEVELOPMENT OF A CONTROL SYSTEM FOR A MILD HYBRID VEHICLE John Ormerod & Peter Fussey Rcardo Consultng Engneers Lmted Abstract: Hybrd vehcles delver the dual benefts of decreased fuel consumpton and exhaust emssons. Modern hybrd vehcles have many controllers for the dfferent systems. For example a powertran may have an engne, electrc motor and gearbox controller (for automatc transmssons), all of whch affect the performance of the vehcle. In addton, there are typcally chasss and vehcle controllers present n the vehcle. Ths can create a control stuaton where more than one controller s able to nfluence the system and there may be conflctng nterests between the controllers. The soluton s an overall supervsory controller, drectng and coordnatng all other controllers, to meet drver demand n response and performance, whlst meetng other goals such as fuel consumpton and emssons. Ths paper descrbes the development of a control system for a mld hybrd vehcle, - MoGen ntellgent Motor Generator. Copyrght 21 IFAC Keywords: 42-Volt; Hybrd; Engne Management; Real-Tme; Control; Fuel Economy. 1. INTRODUCTION Today s challengng requrements for vehcle emssons and energy effcency form the background to ths project. The -MoGen project (ntellgent Motor Generator) s a collaboratve research programme between Rcardo and Valeo to buld a demonstrator vehcle wth the followng characterstcs, reference [1]: C class vehcle Fuel consumpton less than 4 l per 1 km Half Euro IV exhaust emssons -1 kph n less than 1 seconds Excellent drveablty The vehcle layout chosen to fulfll the objectves above used an advanced 1.2 l Hgh Speed Drect Injecton (HSDI) desel engne developng 75kW rated power, a 6kW flywheel mounted electrcal devce (FMED), a Nckel Metal Hydrde (NMH) battery pack and 42v electrcal components. The project demonstrates a fully ntegrated control strategy runnng across a network of controllers, managed by an overall vehcle supervsory controller. Strateges have been modelled and tested n Smulnk, and subsequently coded usng ASCET-SD for mplementaton n the Vehcle and Engne Management Prototypng System (VEMPS) C167 based controller. The mplementatons were valdated aganst the smulaton models performance usng a hardware n the loop approach where possble.
2. SYSTEM DESCRIPTION The nternal combuston engne used was a hghly rated (6kW/ltre) 1.2 ltre HSDI desel engne wth 5 speed manual transmsson. A 42V Flywheel Mounted Electrcal Devce (FMED) was used to start the engne, motor (delverng up to 6kW boost) and to generate up to 6kW. The effcent power generaton allowed a number of hgh power electrcal accessores ncludng brake pump, ar condtonng, power asssted steerng, coolant pump, coolng fans, and desel partculate flter (DPF) heater. 3. SYSTEM ARCHITECTURE The control system layout s shown n Fgure 1. The remanng three control unts manage the 42V systems of the FMED, battery and HVAC. Fuel Injecton EGR VGT Supervsor Controller Thermal Systems Exhaust Aftertreatment 42V Battery FMED EDC7c Rcardo VEMPS A B C BOSCH Valeo 42V HVAC FMED Controller Engne and Flywheel Mounted Electrcal Devce (FMED) Engne Controller 42V thermal system Vehcle Controller Electrc heater n exhaust system Controller Area Network (CAN) Fg. 1: The CAN network used n vehcle. A CAN network connects the followng modules. Battery Controller NMh Battery 42V Heatng, Ventlaton and Ar Con (HVAC) HVAC Controller Vehcle control unt (VCU), VEMPS unt Engne control unt (ECU), Bosch EDC FMED controller Battery controller for NMH 42V batteres HVAC controller (heatng ventlaton and ar condtonng) The followng modules were also used durng development, but not fnal mplementaton: A rapd prototypng controller (ES1) runnng the FMED supervsng strategy durng floatng pont strategy valdaton, pror to portng ths strategy nto VEMPS controller above. A calbraton tool (INCA) for the VCU, va Can Calbraton Protocol (CCP), and ECU va a K-lne connecton Fg. 2: Strategy dstrbuton across control unts The exhaust after-treatment s unusual and merts further descrpton. A desel partculate flter (DPF) s used to trap the soot partcles whch accumulate over tme. Perodcally the flter needs to be cleaned or regenerated. Ths nvolves burnng the soot by rasng the temperature of the flter. Prevous systems have used addtves to lower the temperature requred and have used late fuel njecton to rase the temperature of the exhaust gas. Addtves are not used n the - MoGen vehcle, wth the hgh temperatures beng acheved by a hgh power electrc radatve heater n front of the flter. In addton the engne characterstcs are modfed to rase the exhaust gas temperature and mantan suffcent oxygen n the exhaust system. 4. CONTROL SYSTEM STRATEGIES Hybrd vehcles requre advanced energy management to extract the most beneft from the ncreasngly complex components. The hybrd vehcle confguraton also offers a number of ways of meetng the drver demands. One of the challenges of ths project was to use the supervsor controller to select the best way to satsfy the drver demands, whlst meetng the programme objectves of fuel economy and emsson performance. The engne management control strateges were dstrbuted between the ECU and the VCU to allow the supervsor controller full access to some of the engne management strateges, Fgure 2. For example the exhaust after-treatment occasonally requres the Exhaust Gas Recrculaton (EGR) and Varable Geometry Turbocharger (VGT) to be controlled to rase the exhaust gas temperature. In addton to the EGR and VGT control, the VCU also controls the thermal system and exhaust after-treatment.
The VCU features: 1. Intellgent Energy Management (IEM) - ncludng automatc engne start/stop operaton, FMED boost and battery voltage regulaton, ncludng a cost functon approach to mantan battery SOC. 2. Engne control In addton to EGR and VGT control, the VCU modfes the engne control strategy under 2 condtons; when the FMED s supplyng a varyng generaton power and when the DPF s beng regenerated. 3. Thermal system control The VCU uses a mnmum energy coolng strategy used to select fan and pump speeds, together wth coolant valve poston to mnmse parastc energy losses. 4. Exhaust after treatment The VCU controls the regeneraton of the DPF; decdng when to start regeneraton, managng the regeneraton and t s duraton. 4.1 FMED Strategy The objectve functon s represented as a 3 dmensonal map versus engne speed and load. A specfc objectve functon wth respect to engne power was calculated from ths map and used n the followng manner, summarsed n fgure 3 whch shows the specfc objectve functon versus engne load at a constant engne speed. The control nput s a cost lmt, whch s used to calculate a beneft lmt that s a functon of the cost lmt and the battery/fmed effcency. For engne operatng condtons wth an objectve functon below ths lmt, the engne can be used to generate more electrcty, up to the pont where the new operatng pont reaches the cost lmt. For engne operatng condtons above the beneft lmt, the FMED can be used to reduce the engne load by motorng,.e. the beneft from motorng s greater than the cost nvolved n generatng the electrcty. The FMED strategy (patent pendng) has been ntroduced n an earler paper [2], so an overvew s presented here. The strategy addresses the followng queston: how much power should the FMED generate or motor? Specfc Cost Dervaton of torque values at the cost and beneft lmts Ths queston s subject to two key constrants: The battery state of charge (SOC) should reman constant over a perod of tme The drver torque demand should be met Beneft Lmt Cost Lmt Generate Actual specfc cost vs load Cost curve used by controller Requested powertran torque [Nm] Motor Coupled wth a requrement to mnmse a number of vehcle attrbutes such as fuel consumpton, emssons, nose and vbraton etc. The strategy therefore reduces down to a multvarable optmsaton problem. An objectve functon s ntroduced to address ths problem and s calculated off-lne to allow the FMED usage to be optmsed n real tme. An objectve functon s derved from the normalsed vehcle attrbutes or speces as follows: Objectve Where: _ Functon = ( w ( ) ) 2 Speces Speces normalsed quantty of each speces, calculated as the rate at that nstant, dvded by a reference value w weght for each speces speces: Fuel consumpton, HC, CO, NOx, PM, NVH etc. Fg. 3: FMED strategy The benefts of the strategy compared wth smple rule based strategy were up to 5% reducton n fuel consumpton and up to 7% reducton n exhaust emssons. 4.2Start/Stop strategy A start/stop functon allowng the engne to stop automatcally when the drver selects neutral and restartng when the gear s engaged, was mplemented usng a state machne. A smplfed state machne s shown n fgure 4. The condtons were enhanced wth numerous safety checks to: avod unwanted engne starts preserve the electrcal system, especally mportant wth potentally large power demands wth the engne off from the electrcal HVAC The objectve functon gves rse to a generalsed concept of cost and beneft, a cost beng an ncrease n objectve functon and a beneft beng a reducton.
clock Stall Engne On Key Crank Engne Crank Engne off Cannot auto start Door Open Into Gear Key Off Into Neutral Engne paused Can auto start Key Off DEVELOPMENT TOOLS SOFTWARE HARDWARE Strategy development Personal Computer Matlab/Smulnk TM Key On Engne off Start wth key Prototype Codng Floatng pont Prototype Codng Fxed pont Rapd Prototypng Controller (ES1) Fg. 4: Stop/Start state machne ASCET-SD Producton Codng Fxed pont Producton Controller (eg. c167) 4.3 Coolng STRATEGY The control strategy has closed and open loop components. The control sgnal s specfed as a coolng performance request, so that when the engne s too hot the controller asks for more coolng performance. The closed loop controller reacts to engne coolant temperature. The open loop controller takes the engne speed and load and based on an expermentally determned map, gves the mnmum coolng performance to protect the engne. The optmum pump and fan speed for a gven coolng performance have already been determned off-lne. Hence the thermal system s operated to mnmse the parastc losses caused by the system. 4.4 DPF Regeneraton strategy The strategy has two stages, before and durng DPF regeneraton. Before DPF regeneraton The strategy establshes when to regenerate, based on measurement of soot loadng, and an nternal model of soot generaton n the engne. Once the DPF requres regeneraton, the strategy establshes when condtons are sutable for runnng the heater. Durng DPF regeneraton The VCU determnes when the engne s at a sutable operatng condton to start the DPF regeneraton, wth respect to the exhaust gas temperature and electrcal power avalable. The VCU then controls the heater power va a PWM sgnal and modfes the engne control to ncrease the exhaust gas temperature. The strategy closes the EGR valve and then manages both the ntake throttle to reduce mass flow rate and the VNT to ensure that the drver torque demand s unaffected. Fg. 5: Controller Development The development process s summarsed n Fgure 5. An Integrated powertran model (IPT) model of the full vehcle ncludng all key sub-systems was used to develop the VCU strateges, as descrbed n [2]. The strateges were ntally developed and valdated n a Smulnk smulaton. The strategy was transferred to Ascet-SD where t was combned wth the necessary communcaton nterfaces, as descrbed n fgure 6. Ascet-SD was used to automatcally generate the control code for the supervsor controller processors used n the vehcle, ether the ES1 or the VEMPS Semens C167 based controller. CAN Comms Strategy I/O Drvers Fg. 6: Ascet-SD mplementaton of VCU 5. RESULTS CAN Calbraton Protocol (CCP) The results shown below have been selected to llustrate the development process. 5.1. Smulaton The IPT model was used to develop the energy management strategy. The example shown n fgure 7 demonstrates the automatc start/stop operaton modellng nteractons between the engne, battery and DPF heater.
Engne Speed [rev/mn] 3 25 2 15 1 5 Engne suffcently warm Stop Start Enabled Stop/Start Operaton 2 4 6 8 1 12 Tme [s] DPF heater on Stop Start Dsabled Many of the strateges were proven on a engne dynamometer.the example n fgure 9 shows the command from the VCU to the FMED controller requestng a torque boost n response to a torque defct between the drver s demand and the actual torque of suppled by the engne. The FMED was not avalable n the engne dyno, however these tests allowed the strategy to be checked n the absence of the electrcal hardware. A second example of testng on the engne dynamometer s shown n fgure 1, where the DPF s beng regenerated at engne dle condtons. 7. -MoGen Regeneraton at dle (5.5kW heatng) - Temperatures 14. Fg. 7: IPT smulaton results over the NEDC drve cycle 6. 12. 5.2. Embedded Software The VCU strategy was transferred from Smulnk to Ascet-SD. The transfer was verfed by testng off-lne usng nputs to the Smulnk controller, fed nto the Ascet controller. The results of a the transfer of the pedal mappng s shown n fgure 8. Temperature (degc) 5. 4. 3. 2. 1. 8. 6. 4. Current (A), Voltage (V) 1. 2. 12 1 8 Offlne test for Powertran Torque Demand.. 1 2 3 4 5 6 7 8 9 Tme (s) Pre DPF temperature Post DPF temperature Heater Current Battery Voltage Torque [Nm] 6 4 2-2 -4 2 4 6 8 1 12 Tme [s] Smulnk ASCET Fg. 8: Offlne test of Ascet strategy Fg. 1: Engne dyno testng of Aftertreatment strategy The strateges were progressed nto the vehcle: Fgure 11 shows the FMED boost functon above operatng n the vehcle. Fgure 12 shows the stop start strategy operatng n the vehcle FMED Boost Functon test at 15 rev/mn 14 12 1 Torque [Nm] 8 6 4 2 2 4 6 8 1 12 14 16 Tme [s] Actual Engne Torque (Estmated) Engne Torque Demand FMED Torque Demand Fg. 9: Engne dyno testng of VCU strategy
Boost enabled FMED Boost Drver Torque Demand Actual Engne Torque 6. CONCLUSIONS The -MoGen programme has demonstrated the ntegraton of many complex strateges nto a coordnated network of control unts. The IPT smulatons have enabled up-front development of the control strateges and have carred forward nto the rapd prototypng tools and technques. The realsaton of full vehcle demonstrator hghlghts the benefts of hands on experence also the potental complexty of future vehcles. The vehcle s capable of meetng the targets of a 3% fuel consumpton reducton, half Euro IV emssons and real world drveablty wth a fully ntegrated control system. Fg. 11: Vehcle testng of VCU strategy Demonstraton of Stop/Start strategy n vehcle 25 2 15 1 5 1 3 5 7 9 11-5 Tme [s] Engne Speed [rev/mn] Flag_Neutral Fg. 12: Vehcle testng of stop/start strategy 5.3. Control Hardware Development 7. ACKNOWLEDGMENTS The author of ths paper would lke to thank the drectors of Rcardo and Valeo for permsson to publsh ths paper. 8. REFERENCES [1] R L Gordon, A Systems Approach to the Mld Hybrd Powertran, Rcardo Internatonal Conference 21. [2] P M Fussey, C L Goodfellow, K K Oversby, B C Porter and J C Wheals, An Integrated Powertran (IPT) model Stage 2: Systems Integraton, Supervsory Control and Smulaton of Emssons Control Technology. SAE 21-1-938 JOrmerod@Rcardo.com PMFussey@Rcardo.com 9. CONTACT As wth many projects, the control strateges are only part of the story. Some detals of the control system hardware have been mentoned n earler sectons and here the mplementaton s dscussed brefly. CAN network - Implementaton nvolved a large amount of work to ensure that the systems were all compatble and worked n a robust manner. Ths ncluded some low level modfcatons to the CAN modules n both the VEMPS and the FMED controllers. FMED - EMC analyss and sheldng were necessary to prevent Electromagnetc Interference affectng other modules. Coolng Electrcal components such as the FMED, the NMH 42V batteres, the DC/DC converter, and the FMED controller, requred coolng ether electrcally or by ncluson n the coolng crcut.
1. DEFINITIONS, ACRONYMS, ABBREVIATIONS ASCET: AMT: CAN: CL: DAC: DoE: DPF: ECU: EDC7: Programmng language for embedded software Automated Manual Transmsson Controller Area Network Closed Loop control Dgtal to Analogue Converter Desgn of Experments Desel Partculate Flter Engne Control Unt Bosch Engne Controller INCA: IPT: NOx: OL: PTD: PWM: SOC: VCU: Calbraton tool Integrated Powertran Model Ntrous Oxde exhaust emssons Open Loop control Powertran Torque Demand Pulse Wdth Modulated State Of Charge Vehcle Control Unt VEMPS: Vehcle and Engne Management Prototypng System EGR: Exhaust Gas Recrculaton VNT: Varable Nozzle Turbne FMED: Flywheel Mounted Electrcal Devce HVAC: Heatng, Ventlaton and Ar Condtonng HSDI: Hgh Speed Drect Injecton -MoGen: Intellgent Motor Generator I/O: Input/Output