ICE²: an integrated software to accelerate your calibration process Generating excellence
D2T calibration expertise Calibration - Diesel Engines Calibration Engine tuning Model Based Simulation on test bench Pollutant Optimization / consumption / noise Particulate Filter Performances Cold Start Regulation (Boost Pressure, EGR) EOBD, electrical and functional diagnostic, safety Hot, cold, altitude conditions Drivability 2
D2T calibration expertise Calibration - Gasoline Engines Our Strengths Calibration work on engine and on vehicle Dedicated engineers for calibration work: Basic calibration (air load, torque structure, power optimization) Startability at various temperatures Drivability Knock detection and regulation, OBD (catalyst diagnosis, misfire detection) Safety (torque and components monitoring) Lambda regulation and emissions optimization for Euro 6 standard. Calibration work: On engine (at D2T Testing Centre) On vehicle (i.e at RENAULT Sport facilities) 3
MBC approach Model based calibration Market requires more and more efficient calibration methodologies for engines to meet industrial needs : Meet simultaneously regulation and customer requirements Shorten test duration and accelerate optimization phase Use test data for several applications of the same engine Take into account robustness Take into account interaction between ICE and its surrounding 4
MBC approach Model based calibration Why do we need a MBC approach? Emissions regulation and engine performance s seek, leads to a higher complexity of calibration work Diesel engines combustion parameters to be tuned increased strongly with multiple injection strategies and advanced air path (TGV HP/LP, EGR HP/LP, Air throttle, etc) Gasoline engines complexity as well have become more complex with direct fuel injection, VVA systems (VVT, VVL), turbocharger and EGR Analytical models cannot represent complex phenomena (i.e NOx, PM emissions), and thus calibration with simulation is today still a long path challenge 5 Experimental testing needed
MBC approach Model based calibration Experiments, but at what price? A standard sweep approach is not compatible 8 factors with 4 different levels makes a total of 65536 experiments Need of models based on experimental results to limit the number of test Fitting polynomial models depending on 8 factors require around 100 experiments DOE/MBC approach is completely adapted to solve the optimization problem 6
MBC approach Work Flow (1/2) Work flow for cycle calibration of MBC local approach example Cycle analysis Selection of pertinent operating point Domain definition Experiment definition 7
MBC approach Work Flow (2/2) Work flow for cycle calibration of MBC local approach Testing Modeling Optimization Smoothing of maps 8
MBC approach Methodologies So, calibration methodologies are now based on DoE approaches But different methods to do it Local Approach Mixed Approach Global Approach Model of engine responses Local models Local models Global models Optimization and Local optimization + Direct optimization Direct optimization of maps building smoothing of maps maps Need of a tool to summarize and integrate all methods and data 9
Overview Model based calibration tool : ICE² (Integrated Calibration Environment for Internal Combustion Engine Design of Experiment, Modeling, and Optimization tool in a same software Possibility of a cycle based calibration Engine calibration oriented Innovative map smoothing tool Direct map optimization Many features coming soon with further releases (after treatment efficiency, etc) 10
ICE² goal What for? In order to Be more efficient Reduce, rationalize and enhance the experimental phases Capitalize know-how of IFPEN/D2T in terms of advanced methodologies and innovative algorithms How? Assisted by computer for all or part of the calibration tasks For whom? Calibration technicians and engineers 11
ICE² main features and innovations Study of normalized cycles on speed/load operating points Discretization (divide) into operation areas (strategies) then into operating points Time and energy weighting of operating points Management of driving cycle with hot and cold start Experimental approach using model-based calibration Tests design in a non-hypercubic domain is managed Statistical modeling of engine responses, including kriging techniques Constrained optimization settings, with robustness analysis Maps deformation Maps smoothing and direct optimization techniques Calculation of a map smoothing index Simultaneous optimization of hot and cold maps actuator engine actuator Estimation of emissions on driving cycle Consistency with automation system, ECU interface and post processing 12
Organization: Application Explorer All data are in 3 main nodes into the project Theoretical or experimental data Raw or digital data CYCLES node For driving cycle analysis Engine Outputs node For validated data from engine test bench Maps node For smoothing maps and optimization 13
Organization: Editors and views The editors and displays can be organized as tabs can interact if they are from the same data into the main Engine Outputs node can be displayed together 14
Organization: Additional views zone Synthesis panel Allows the user to place the progress of project tasks form of Workflow For exportable form of image file via File menu / Export for insertion into a report or a presentation 15
ICE² work flow Cycles node : Importation of cycle trace 16
ICE² work flow Cycles node : Distribution of operating points into operating areas constant engine control mode 17
ICE² work flow Cycle nodes : Weighting of operating points 18
ICE² work flow Cycle nodes : Validation of cycle discretization (with engine photo data) 19
ICE² work flow Engine outputs node : domain modeling Example local of air path domain for different injection parameters sets 20
ICE² work flow Engine outputs node : design of experiment into the domain D-optimal design or space filling Example 6 parameters space filling DoE (pairwise projection) 21
ICE² work flow Engine outputs node : response modeling Use of polynomial or kriging models o Modeling statistical results o Graphical determination of model quality 22
ICE² work flow Engine outputs node : local optimization (for local approach) Minimization of one objective with respect to constraints on other engine responses or factors (domain) Take into account dispersion of factors 23
ICE² work flow Maps node : smoothing between local optima to obtain final maps (local approach) smoothing principle smoothing results global regularity index (on each control map) cumulative results 24
ICE² work flow Maps node : direct optimization of maps (mixed approach) 1. parameterization of control maps (bilinear or Lolimot) 2. definition of objective and constraints 3. optimization process 4. analysis of results 3D comparative view of initial and final control maps Instantaneous and cumulated emissions 25
Local approach overview Normalized Cycle From the initial calibration settings and results to the final optimized ones New maps validation Cycle discretization and weighting into N operating points (OP) Domain definition (on NxOP) N local optimizations Optima integration and map smoothing Optimization, and robustness (on NxOP) Test plan (on NxOP) Modeling (on NxOP) 26 Testing
Mixed approach overview Normalized Cycle Cycle discretization and weighting into N operating points (OP) Domain definition (on NxOP) From the initial calibration settings and results to the final optimized ones Direct map optimization with cycle objectives New maps validation Cycle driven optimization, smoothing and robustness at the same time Same process as Local approach Test plan (on NxOP) Modeling (on NxOP) 27 Testing
Conclusion Engine calibration oriented Fully integrated tool and easy to use Standard and kriging modeling techniques Maps smoothing tool and visualizations Direct map optimization (cycle driven) D2T-IFPEN know-how inside Further innovative developments and improvements on going The right tool to accelerate your calibration process and ensure meeting quickly today s and tomorrow s new requirements 28