The influence of advanced boosting on transient NOx control in Light Vehicle Diesel engines

Transcription

1 BMEP [bar] The influence of advanced boosting on transient NOx control in Light Vehicle Diesel engines E. Bouvier 1, D. Kihas 2 J.-S. Roux 1, S. Vankayala 3, D. Jeckel 1, C. Rivière 1, M. Uchanski 4 1: Honeywell Turbo Technologies, Thaon-Les-Vosges, France 2: Honeywell Automotive Software, North Vancouver, Canada 3: Honeywell Technology Solutions, Bangalore 4: Honeywell Technologies Sarl, Rolle, Switzerland Abstract: 2017 and 2020 will represent major milestones for the light vehicle Diesel engine in Europe and other regions that are committed to WLTP. The Diesel engines position as the most efficient prime mover and the need for it to be included in a balanced and aggressive CO2 roadmap is not really disputed. Diesel however faces significant challenges in WLTC / RDE to deliver low real world engine out and tail pipe out NOx at a competitive cost. This paper explores how advanced Boosting Technologies can be used to enable Fuel Injection, Combustion, Aftertreatment and Control systems to significantly impact both Engine Out and Tail-pipe Out NOx. It will provide a baseline and some limits of what conventional Mono & TwoStage solutions can provide today and go on to show how these systems could evolve to help reduce NOx emission and how they could co-exist with electrical boosting options in the context of Mild Hybrid architectures. Keywords: Boosting, Optimal Control, Model Predictive Control, Simulation, RDE 1. Introduction Internal Combustion Engine systems for automotive applications are becoming increasingly complex due to: Higher complexity of Hardware Increasing number of required sensors and actuators Diversity of architectures for markets and applications Certification now including WLTC/RDE More and more stringent regulations Complexity of the development process that requires hand in hand iterations of hardware solutions and control software solutions in order to meet the desired performance parameters. The present paper illustrates Honeywell design & simulation initiatives that are leading toward suitable engine architecture and control strategy, achieving desired transient NOx behavior over a wide engine operating range. It will mainly focus on development challenges and methodology. Note about graph format: In this introduction part, graphs will show ranges of LV diesel engine data with mono-stage VNT TM turbocharger. A representative sample of Euro6b engines will be compared with a representative sample of likely Euro6d engine performance. Minimum and maximum values of each sample data will be plotted. Most of these graphs will refer to full load operation of the engine. 1.1 RDE impact on engine For Euro6d regulation, engine manufacturers are developing a high diversity of strategies to comply with Real Drive Emissions. It means significant consequences on engine sizing, then max BMEP may still be increased (continuing downsizing trend on some applications), whereas most of engines would show a stable or decreasing BMEP (rightsizing and upsizing trends). As shown on Figure 1, this leads to 50% wider range of max BMEP in Euro6d vs Euro6b engines Engine Speed [rpm] Eu6b Likely Eu6d 80kW/L 40kW/L Figure 1: Brake Mean Effective Pressure in full load To comply with Euro6d regulation, significant EGR rates in full load are expected, up to 10% EGR: such practice was almost unknown in LV diesel engines compliant with Euro6b regulation (cf. Figure 2). Page 1/10

2 Compressor inlet temperature in full load [ C] 1.3 Steady challenge for T/C To control emissions, engine developers are working on both engine out emissions and tailpipe emissions increasing EGR rate, or upgrading post-treatment. More and more engine architectures will be longroute EGR and SCR or LNT/SCR to comply with Euro6d -and beyond- regulations. One consequence of Long-Route EGR on turbocharger, see Figure 5, is increased compressor inlet temperature, up to 10 C in full load operation Figure 2: EGR rates on full load 1.2 Depollution zone in engine map The Figure 3, showing C-class vehicle 6 gears DCT simulation from Powertech, and the sketch at Figure 4 both illustrate the extension of depollution zone in engine map Eu6d RDE/RTS95 versus Euro6b. NEDC [1] Engine Speed [rpm] Eu6b Likely Eu6d Figure 5: Compressor inlet temperature in full load Despite significant efforts from post-treatment suppliers to contain drawbacks of post-treatment upgrade, an increased pressure drop in exhaust line -and therefore a higher turbine outlet pressure- by 150 to 200 mb in average is expected as shown on Figure 6 [2]. Figure 3: Eu6d/RTS95 EGR map BMEP Full load Figure 6: Turbine outlet pressure in full load Engine speed Figure 4: Depollution zones vs cycles in engine map As can be seen in Figure 7, wider BMEP range along with higher EGR rates are leading to Euro6d having almost twice as wide a range of Engine intake pressure as Euro6b Page 2/10

3 Compressor outlet temperature [ C] Turbine inlet pressure [bar g] Engine intake pressure [bar g] Eu6b Likely Eu6d Engine speed [rpm] Figure 7: Engine intake pressure in full load As a consequence of previously mentioned parameters, turbine inlet pressure, on Figure 8, and compressor outlet temperature on Figure 9, are significantly more severe on Euro6d engines than Euro6b ones, pushing design limits of both compressor and turbine stages. 4 severe boundary conditions, with OEM expectation to deliver game-changing performances. 1.4 Reliability To compare these Euro6d boundary conditions with turbocharger limits, following graphs show operating points in compressor and turbine maps. As it can be seen on Figure 10, Euro6d compressor requirement in full load is wider versus Euro6b: Flow range with operating points approaching choke line: enabler to manage over-speed risk is Honeywell speed sensor technology Pressure ratio range: 50% wider Constraint on surge operation is still highly present with engine points beyond gas stand surge line Eu6b Likely Eu6d Engine speed [rpm] Figure 8: Turbine inlet pressure in full load 250 Figure 10: Full load operation in compressor map Figure 11 suggests that turbine requirement is also severely increased with 50% wider pressure ratio range in full load operations in our LV diesel engine mono-stage VNT TM representative sample Eu6b Likely Eu6d Engine speed [rpm] Figure 9: Compressor outlet temperature in full load As a conclusion, Euro6d regulation impact on turbocharger is a rise to unprecedently wide and Figure 11: Full load operation in turbine map Page 3/10

4 Compressor efficiency [-] 1.5 Emission regulation impact on transient turbocharger performance Within Euro6b boundary conditions on LV diesel engines, standard strategy was to use EGR mainly in steady state part-load, and avoid it in transient to maintain a good time to torque performance. To comply with Real Drive Emissions, usage of EGR in transient including full-load is expected to be generalized. For the same reason as in steady state (see 1.3 and 1.4), it will mean more work requested from turbocharger in Euro6d vs Euro6b, then highest engine sensitivity to turbocharger transient performance. 1.6 Impact of catalyst light-off constraint on turbocharger During engine warmup, depollution capacity of engine depends highly on the gas temperature in exhaust line, so that after-treatment works with optimum efficiency. After-treatment is generally located downstream of the turbocharger, then constrain for it is to absorb minimum heat. To comply with RDE, turbocharger will need to: be more efficient, improve its transient response, enable early catalyst light-off during engine warmup be reliable in unexplored pressure conditions be controlled through robust strategies 2. Honeywell boosting path 2.1 Developments overview As shown on Figure 12, Honeywell has developed a wide range of boosting solutions to cope with new emission regulations. Among them, VNT TM still represents the foundation serving with adaptations- both Mono-Stage and Multi-Stage applications. As mentioned in past papers, other concepts / architectures are also investigated. 1.7 Need for emission simulation capability Increasing complexity of boosting and engine architecture are simultaneously expected with faster turbocharger development time from turbocharger suppliers. To enable it, simulation capability is key to fully understand engine constraints and differentiate turbocharger offering. 2 will describe current simulation capabilities and will discuss how to address the turbocharger impact on engine emissions, in full and part load operations, in steady-state as well as in transient conditions. 1.8 Need for robust and optimized controllers Real Drive Emission regulation will likely have a significant impact on number of actuators to be controlled and diagnosed by engine ECU. EGR control strategy impact on emissions will be briefly presented in 2 to justify the new need for robust and fully optimized controllers in both simulation and real world. Then in 3 Honeywell multi-variable controller solution OnRAMP TM built from Honeywell s advanced control technology experience will be presented as investigation route to address control challenge [3]. To sum up this introduction, high diversity of engine strategies to comply with Euro6d and Real Drive Emissions are resulting in extremely severe boundary conditions for the turbocharger. Figure 12: Boosting solutions 2.2 Last VNT TM : GTE overview The new VNT TM generation GTE is already a significant improvement versus previous generation GTD. As shown on Figure 13, compressor is showing map width increase along with close to surge efficiency improvement. Compressor efficiency eyebrows Compressor corrected flow [g/s] Figure 13: Compressor map GTE vs GTD Similarly, see Figure 14, turbine has also improved slightly map width. However, the new turbine stage combined with Honeywell optimized bearing Zultra GTD GTE Page 4/10

5 shows significant efficiency improvement at low flow, critical to low-end Steady and transient performance [4], [5],[6],[7]. Figure 14: Turbine GTE vs GTD 2.3 Simulation approach: GTE vs GTD example The need to accelerate development at a reasonable cost has lead Honeywell to make crucial investments in numerical toolchain from component up to vehicle level. Powertrain simulation example in Figures 15 & 16- does not only allow to evaluate new concept value but also to correctly define targets and optimize concepts and aerodynamics [5]. Figure 16: Example of GT-Power / GT-Drive models As shown on Figure 17, the overall GTE compressor, turbine and bearing benefits could therefore be evaluated at engine / vehicle in Eu6b context Figure 17: GTE Overall Performance benefit GTE represents a step forward in terms of CO2/ performance and will enter production stage in In order to further help OEMs reach Eu6d/RDE compliance, development activities around GTE successor GTF started in GTF development goes hand in hand with new methodology that will be presented in the following section. 2.4 Simulation challenges Figure 15: Example of FRM GT-Power model NOx emissions estimation As it can be seen on Figure 17, the radar chart is only showing CO2 estimation. GTE had indeed been evaluated at that time with same EGR strategy as GTD. GTE can therefore be seen at almost iso engine out NOx vs GTD. Despite emission simulation absolute accuracy concern at OEM side, relative results providing magnitude, ranking and trends are frequently used at Tier I [8],[9]. In 2015, Honeywell decided to start feasibility study of NOx simulation in GT-Power. Page 5/10

6 The first step has been to look at load step and sweep EGR valve position. Those results summarized on Figure 19 show very highly consistent magnitude despite the diverse engine and cycles considered. They overall confirm that turbocharger design (Inertia / efficiency) & control optimizations can reduce engine out NOx up to 30% on transient events and up to 8% on Real Drive cycles such as ARTEMIS. Figure 18: NOx/CO2Trade-off on a load step Figure 20: NOx reduction, [10],[11] NOx control challenge on cycle The NOx / CO2 trade-off obtained through parametric sweeps (EGR, VNT TM ) presented before is good enough to understand relative positioning of various turbo technologies however it can t be applied to a driving cycle simulation due to the number of simulations to be run. Several simulations studies have therefore been conducted with AVL Graz and TU Dresden in order to confirm magnitude found at Honeywell side and identify state-of-the-art approaches. In Figure18, an example of GT-Power Load Step simulation conducted by TU Dresden explaining the mechanism. The turbo B delivering faster boost than A shows benefit on NOx only through an optimized EGR recalibration. NOx / EGR control logic need appears pretty obvious however the complexity is rising hugely when moving from Mono-Stage to Multi-Stage (HP/LP VNT position, EGR valve, TBV,...).or electrified boosting (e-boost / turbo split). The main challenges of NOx control is inclusion of Multi-parametric optimization and Multivariable optimal control with constraints handling as early as possible in the development cycle (e.g. at powertrain simulation stages). Figure 20 shows the adoption of 1D tool such as GT- Power but as well the trend to connect Controls and engine simulation usually through Co-Simulation. Multiple paths exist and will be explored in 2016 to bring internal capabilities. Honeywell sees an opportunity to include its own software suite OnRAMP TM in robust NOx control plan. It will be detailed in next section. Figure 19: NOx Transient control example Similar optimization studies have been applied to complete driving cycle using different toolchain (GT- Power and MoBEO). Figure 21: Honeywell simulation investigation path Page 6/10

7 3. Honeywell software developments The OnRAMP TM Design Suite provides a systematic approach to control design process starting from system modeling [3], [12], [13], [25]. It makes the modeling process more efficient and less dependent on the skill level of the engineer. The tool provides environments for building engine scheme out of engine elements library, design of experiment, data management, data analysis and fully automated model calibration based on experimental data and component data such as turbocharger maps. Further, the tool has model validation features and both the local performance of the individual components and the global input-output behavior of overall connected model can be evaluated. Notably, the system should transparently handle the complex multivariable interactions that often arise when a new actuator is added. Interchangeable between simulation and dynamometer, so that the control system can be coupled with simulation models for virtual screening. Constraint management, in order to avoid damage to costly prototype hardware and reduce the time and care which must be taken to develop setpoint tables. Production readiness, so that the prototype control system can gracefully merge into the production workflow if desired. Figure 22: OnRAMP Component Identification Toolset The models and controllers produced by OnRAMP TM can be exported as Simulink/Matlab software components, as the tool partially relies on elements of Simulink/Matlab infrastructure, and in that way enables software level interfacing. 3.1 Control Performance Requirements and Hardware Selection Hardware selection is a combination of paper studies, simulation, steady state open-loop experimentation, and drive cycle tests for drivability and transient emissions. Requirements for a control system to be considered to support the Hardware Selection include: Fast turn-around and low test bed time, which allow for more hardware iterations in a given calendar time. Near optimal performance, so that hardware performance is not masked by controls performance. Flexible with respect to hardware changes, for example to allow comparison of single and dual turbocharger options or of high pressure EGR vs. dual loop EGR. 3.2 OnRAMP Control Design Process OnRAMP control design software tools provides a systematic MPC based control design process [3]. It makes the control design and tuning process more efficient and less dependent on the skill level of the developer than most traditional controls approaches. The design process was developed with the goal of integration into industrial practice [3]. Careful consideration was given to existing development processes for industrial engine control and how the resulting control function interfaces in the hierarchical software structure [21], [26]. A typical sequence of MPC control design steps supported by the mentioned software package is listed below [12], [13], [25]: 1. Definition of the model structure based on the engine configuration and physical laws 2. Design and execution of test cell experiment in order to acquire the data for model calibration (3 test cell days) 3. Automated model calibration based on the obtained experimental data and available component data (e.g. turbocharger maps) 4. Controller configuration definition that includes specification of inputs and outputs 5. Definition of the control optimization problem, the cost function in terms of setpoints, constraints and their relative importance factors 6. Synthesis of the controller dataset that is deployed with the controller on rapid prototyping system or engine control unit 7. Controller fine tuning calibration in the test cell (Typically 3 test cell days or 1 day of SIL) Page 7/10

8 Model Structure (1) (2, 3) (4, 5) Figure 23: Modeling, Control & Deployment Workflow (6, 7) It typically takes six days of test cell time to reach a near-production calibration on a new hardware configuration. Other work, including project organization, ECU setup, and desktop calibration (steps 1, 3, 4, 5, 6) typically takes another 10 days on the calendar for the first calibration. For subsequent calibrations, non-recurring work is removed and recurring work becomes routine, so time per iteration can substantially improve. The methodology is expected even more efficient when utilized in simulation Model Calibration The model calibration uses either the measurement data or as in this work simulation data from a high fidelity model to fit the mean-value model parameters so that the behavior of the true engine is closely matched. The model calibration step is carried out in two major steps steady-state matching and transient identification. Steady-state matching includes: 1) steady-state component-level identification that provides reasonable starting points for the global steady-state identification, and 2) the global steady state identification that reconciles the components in order to achieve very high accuracy of the overall model. Finally, transient identification step identifies remaining dynamic parameters. The tool uses an optimization algorithm based on nonlinear least squares for all stages of model identification [13] Controller Structure The controller structure calibrated by OnRAMP design process includes MPC feedback controller with Kalman Filter used as observer, feedforward static look-up tables, constraints manipulation element [12], [22], [23], [25], [26], [27]. The complete structure is scheduled on several selected variables that describe the changes of behavior (including nonlinearities) of the targeted system over the operating range [21], [28]. The main role of feedforward control is to show a correct steady-state actuator position to the controller during heavy transients across a wide set of operating points where the model uncertainty is very significant. The main role of feedback control is to provide steady-state offset-free tracking, transient behavior improvements, constraints handling and disturbance rejection. While feedforward control can react faster during heavy transients and provide initial actuator positions, feedback will ensure adjustments of actuator positions to the desired values. The combination of these two can be governed by feedback range concept to ensure that actuators signals are constrained to lie within specified constant range from feedforward signal and in that way provide safe control actions in broad set of highly nonlinear engine applications with large transients [27] Model Predictive Control Algorithm Model-based Predictive Control (MPC) is a widely accepted method for designing optimal multivariable control. The main advantage of MPC over the other control techniques, such as H2, LQG, H, musynthesis (e.g. [29]), is its systematic approach to regulate process with various time-varying constraints on the system inputs, internal states, and outputs. MPC algorithm computes an optimal future trajectory of selected system inputs so that the plant behavior is as close as possible to the requirements. The control goals are expressed by a cost function that contains weighted combination of terms for actuator movements, tracking error and soft constraints. The cost function is minimized by the MPC algorithm over a selected prediction horizon. More on MPC can be found in the following literature [12], [14], [15], [16], [17], [18], [19], [20] Automatic Tuning Algorithm The relationship between weights in the cost function of the MPC problem and the desired robustness and performance conditions can be established through frequency domain analysis of a type of transformed MPC control problem. In this way, it is possible to derive robust stability condition using small gain argument applied to additive uncertainty model. The algorithm manipulates multipliers of weights in the cost function such that the derived robust stability condition is satisfied. The user defined requirements on robustness and bandwidth affect the shape of the derived robust stability condition and in that way the resulting weights of MPC cost function are influenced by the specified requirements for robustness and bandwidth [24], [30]. Page 8/10

9 3.3 OnRAMP and GT-Power Co-Simulation for NOx control Pilot projects will take place during 2016, covering mono-stage in the first case and then exploring TwoStage including e-charger TM options. The series of simulation tests will allow for comparison of OnRAMP TM controller robustness and performance vs standard GT-Power controllers. The interfacing of OnRAMP TM and GT-Power can be accomplished using Simulink/Matlab software as both software tools can either deploy controllers and models to Simulink environment or, alternatively, communication between OnRAMP TM controller and GT Power model can be established using GT- Power harness component within Simulink environment. Performance of both of these methods is currently being investigated as part of the project. 3.4 OnRAMP for on engine NOx control As discussed, the flexibility of OnRAMP TM modelling and controller development enables exploration of overall engine control performance through series of iterations in simulations and in this way can support solution of complex control problems such as NOx control. The combination of the tool flexibility and its ability to quickly and economically attain good control performance levels positions OnRAMP TM as a good choice for projects requiring quick hardware operation for performance evaluation. The described co-simulation methodology, planned for completion by the end of 2016, will shorted the time toward and increase the likelihood of success for on-engine investigation planned at Honeywell / PEH with the goal of demonstrating potential of the final hardware solutions. 4. Conclusion Real Driving Emissions and Euro 6d represent a significant challenge for the entire Automotive Industry including turbocharger suppliers. It does not only need new boosting products or architectures to address more severe boundary conditions such as EGR rate and exhaust backpressure increase but also would benefit from simulation and robust control development to evaluate -at least in a relative way- NOx emissions and then optimize the entire powertrain at system level. Several studies at Honeywell and its partners sides have been conducted and could demonstrate in this paper that: NOx simulation -even if needing detailed validation on engine and standardization - seems to be soon possible at Honeywell side Turbocharger realistic improvements in efficiency / inertia combined with adapted control strategy could enable up to 30% NOx reduction during transient event and up to 8% on Real Drive cycles Control strategy / calibration / boosting architecture must be optimized together to minimize engine out NOx Multiple routes to optimize control both at simulation and test stages have been identified including Honeywell OnRAMP TM robust control solution and are planned to be explored on Acknowledgement The authors would like to extend their thanks to all Honeywell Engineers and Technicians who are contributing to the development of new generation of turbochargers and controls and also to Mr Pautasso, Mr Servetto, Mr Mustafaj, and Mr Olivaud from POWERTECH Engineering, to Dr Walter, Dr Werner and Pr Zellbeck from TU Dresden and to Mr Valchev, Mr Schüßler and Dr Schiffbaenker from AVL Graz 6. References [1] Krüger, Maier, Naber, Scherer, Schumacher, Strobel, Robert Bosch GmbH Diesel Systems, Emission Optimization of Diesel Passenger Cars to Fulfill "Real Driving Emission (RDE)" Requirements, 24th Aachen Colloquium Automobile and Engine Technology [2] Boger, Bhargava, George, Heibel, Nickerson, Rose, St.Clair, Tanner, Vileno, Corning, Advanced High Porosity Filter Technologies to Meet Future Gasoline and Diesel Regulations, 14th Hyundai-Kia International, Powertrain Conference, [3] Honeywell OnRAMP website: [4] P.Barthelet, J.Mailfert, E.Bouvier, C.Riviere, N.Morand Honeywell Turbo Technologies: The smallest Honeywell VNT TM Turbo design for CO2 reduction. 4ème Congrès International Diesel, SIA Rouen, [5] J-S. Roux, M. Marques, D. Armand, C. Wilkins, S. Pees, Honeywell Turbo Technologies, The next generation of small gasoline turbochargers to meet EU6.2, SIA Versailles, [6] P. Barthelet, N. Morand, B. Chammas, L. Toussaint, L. Sausse, C. Riviere, Honeywell Turbo Technologies, The new family of VNT TM turbocharger developed by Honeywell Turbo Technologies for Eu6 and beyond, 18. ATK Dresden, [7] P. Davies, D. Jeckel, G. Agnew, G. Figura, 25 Years of VNT TM Technology Past, Present & Future, P. Barthelet, 20. ATK Dresden, [8] F. Huck, PSA Peugeot Citroën, Predictive Combustion on Diesel Engine for Heat Rejection Determination, GT-SUITE conference Frankfurt, Page 9/10

10 2014. [9] A. Gallone, S. Pizza, A. Capra, M. Rimondi - General Motors Powertrain Europe, D. Bellomare - Powertech Engineering, Study and Application of Predictive DI-Pulse Model For Diesel Combustion in GT-Power, GT-SUITE Conference, Torino, [10] Forissier, Zechmair, Weber. Criddle. Durrieu, Picron, Menegazzi, Surbled, Wu, Valeo Powertrain Systems, The Electric Supercharger, Improved transient behavior and reduced CO2 as well as NOx Emissions at the same time?, 18 ATK Dresden, [11] R. Busch, J. Jennes, J. Müller, Bosch Mahle Turbo Systems GmbH & Co. KG, M. Krüger, D. Naber, H. Kauss, Robert Bosch GmbH, Emission and Fuel Consumption Optimized Turbo Charging of Passenger Car Diesel Engines, Internationales Wiener Motorensymposium, [12] Stewart, G.E., Borrelli, F., Pekar, J., Germann, D., Pachner, D., and Kihas, D., Toward a Systematic Design for Turbocharged Engine Control, in L. del Re, F. Allgower, L. Glielmo, C. Guardiola, Automotive Model Predictive Control, Lecture Notes in Control and Information Science, Springer-Verlag, Berlin-Heidelberg, [13] Pachner, D., Germann, D., and Stewart, G.E., Identification Techniques for Control Oriented Models of Internal Combustion Engines, in Alberer, D., Hjalmarsson, H., del Re, L., Identification for Automotive Systems, Lecture Notes in Control and Information Science, Springer-Verlag, Berlin-Heidelberg, [14] Maciejowski, J.M., Predictive Control with Constraints, Pearson Education Ltd, [15] Bemporad, A., Morari M., Dua, V., Pistikopoulos, E. N., The explicit linear quadratic regulator for constrained systems. Automatica, 38, 3-20, [16] Muske, K. R., Badgwell, T. A., Disturbance modeling for offset-free linear model predictive control, Journal of Process Control 12, , [17] Maeder, U., Morari, M., Offset-free reference tracking with model predictive control, Automatica 46, , [18] Borrelli, F., Baotic, M., Pekar, J., Stewart, G., On the computation of linear model predictive control laws, Automatica 46, , [19] del Re, L., Allgower, F., Glielmo, L., Guardiola, C., Automotive Model Predictive Control, Lecture Notes in Control and Information Science, Springer-Verlag, Berlin- Heidelberg, [20] Qin, S. J., Badgwell, T. A., A survey of industrial model predictive control technology, Control Engineering Practice 11, , [21] Stewart, G.E., and Borrelli, F., System for gain scheduling control, USPTO number , October [22] Khaled, N., Pekar, J., Fuxman, A., Cunningham, M., and Santin, O., Multivariable Control of Dual Loop EGR Diesel Engine with a Variable Geometry Turbo. SAE World Congress, paper , [23] Kim, Y.W., K., Van Nieuwstadt, M., Stewart, G., Pekar, J., Model Predictive Control of DOC Temperature during DPF Regeneration. SAE World Congress, paper , [24] Fan, J., and Stewart, G.E., Patent Automatic tuning method for multivariable model predictive controllers, USPTO number 7,577,483, May [25] Alfieri, V., Pachner, D., Enabling Powertrain Variants through Efficient Controls Development, SAE World Congress, [26] Stewart, G.E., Borrelli, F., A Model Predictive Control Framework for Industrial Turbodiesel Engine Control, 47th IEEE Conference on Decision and Control, Cancun, Mexico, December 9-11, [27] Pekar, J., Stewart, G.E., Kihas, D., Borrelli, F., Patent Method and system for combining feedback and feedforward in model predictive control, USPTO 8,145,329, March [28] Rugh, W., and Shamma, J., Research on gain scheduling. Automatica, , [29] Doyle, J., Francis, B., and Tannenbaum, A., "Feedback Control Theory", [30] Fan, J., Model Predictive Control for Multiple Cross-Directional Processes: Analysis, Tuning, and Implementation, PhD Thesis, University of British Columbia, BMEP: DCT: e-charger TM : ECU: EGR: FRM: HP: LNT: LP: LQG: LV: MPC: NEDC: NMPC: OEM: PEH: PRC: PRT: RDE: SCR: SIL: TBV: T/C: VNT TM : WLTC: 7. Glossary Brake Mean Effective Pressure Dual Clutch Transmission Electrically Driven Compressor Electronic Control Unit Exhaust Gas Recirculation Fast Running Model High Pressure Lean NOx Trap Low Pressure Least Quadratic Gaussian Light Vehicle Model based Predictive Control New European Driving Cycle Non Linear Model Predictive Control Original Equipment Manufacturer Powertrain Engineering Houses Pressure Ratio Compressor Pressure Ratio Turbine Real Drive Emissions Selective Catalytic Reduction Software In the Loop Turbine Bypass Valve Turbocharger Variable Nozzle Turbine Worldwide Light vehicles Test Cycle Page 10/10