Simulation of the Tail-Pipe Emissions fora Heavy Duty Diesel Engine in GT -Power

GT-Suite Users MeetingOct. 20, 2012, Frankfurt/Main

Dr. Joachim WeißDipl.-Ing. Markus Raup

Dr. Thorolf Schatzberger

Dipl.-Ing. Friedrich ForsthuberDipl.-Ing. Thorsten Krenek

Assistant Prof. Dr. Thomas Lauer

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 2

Content

� Introduction

� Simulation model

� Validation Results

� Variation of exhaust system

� Optimization – mean value model

� Summary & outlook

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 3

IntroductionMotivation

� State-of-the-art engines are complex systems with numerous degrees of freedom

� Fuel efficiency is strong demand from the customer

� Emission regulations getting continuously more severe

� Comprehensive model required to study system behavior

� Simulation model as basis for numerical optimization

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 4

IntroductionEngine & Simulation Software

Heavy duty engine for truck and bus application:

� In-line 6-cylinder Diesel engine

� High-pressure common rail injection system with multiple injection pulses

� Two-stage turbo charging with intercooler

� Cooled external EGR

� Selective catalytic reduction (SCR)

� Diesel particulate filter (DPF)

� Diesel oxidation catalyst (DOC)

Simulation Software

� GT-Suite™ for simulation of engine and aftertreatment

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012| Frankfurt/Main | F. Forsthuber | Slide 5

Simulation ModelCombustion Modeling – Spray Model

Source: Hiroyasu et al., SAE-Paper 930612

Spray model• Input: injection rate vs. time

• Spray divided in parcels

• Evaporation and heat release for individual parcel is computed

• Global heat release is summed up

• NOx model based on extended Zeldovich mechanism

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 6

Simulation ModelCombustion Modeling – Resulting Cylinder Pressure

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 7

Simulation ModelAftertreatment Modeling

Flow components (catalyst bricks)

Coupled with reaction kinetics objects

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 8

Simulation ModelAftertreatment Modeling – Reaction Kinetics

Reaction kinetics objects account for:

� Bulk diffusion

� Surface storage

� Gas and surface reactions

Adsorption of ammonia

Desorption of ammonia

Oxidation of ammonia

Standard SCR reaction

Slow SCR reaction

Fast SCR reaction

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 9

Simulation ModelOverall System Modeling

Predictive engine model for fuel consumption and NOx formation

Aftertreatment model with detailed reaction kinetics

Combination of the two parts

required

� One single model with two separated flow circuits

� Boundary conditions of circuits passed through by a control object

� Appropriate timestep for each circuit to provide fast runtimes

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 10

Validation ResultsSteady-State Engine Operating Points – 50% Load

� Comparison between measurements and simulation results of different operating points

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 11

Validation ResultsSteady-State Engine Operating Points – Parameter Variation

Variation of air ratio (Lambda) in one specific operating point

Engine control strategy: Increasing Lambda � decreasing EGR

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 12

Validation ResultsAftertreatment

� Steady-state operating point� Constant exhaust mass flow, composition and temperature� Injection of urea solution (hyperstoichiometric)� NOx conversion and NH3 storage starts� Cut-off of urea injection

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 13

Validation ResultsAftertreatment

� WHTC based transient cycle

� Last section of cycle – urea dosing and NOx conversion start

� Measured raw emissions as input for afterteratment model

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 14

Validation ResultsTransient Cycle

� Combined model with detailed engine and aftertreatment model

� Transient cycle based on the Worldwide Harmonized Transient Cycle (WHTC)

� Fully transient controls for engine and atftertreatment operation analogous to the hardware engine control

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 15

Validation ResultsTransient Cycle – Parameter Variation

� Variation of air ratio (lambda factor) over the entire transient cycle

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012| Frankfurt/Main | F. Forsthuber | Slide 16

Validation ResultsTransient Cycle – Parameter Variation

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012| Frankfurt/Main | F. Forsthuber | Slide 17

Variations of Exhaust System

Base Reduced lengthAir-gap

insulation 1Air-gap

insulation 2

be 100% +0,2% -0,1% +0,1%

NOx 100% -0,3% -0,3% -0,7%

NOx,EOP 100% -14% -13% -19%

tAdBlue 100% -5,1% -4,9% -6,2%

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 18

Optimization – Mean Value Model• One neural network for

each output parameter:• IMEP• FMEP• Volumetric Efficiency• Exhaust temperature• NO emissions• NO2 emissions

• 15,000 steady-statesimulations for neuralnetwork training

Runtime reduction:15 hrs → 45 min for a 1800 s transient cycle

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012| Frankfurt/Main | F. Forsthuber | Slide 19

Optimization – Mean Value Model

� Design of Experiments

� Full factorial

� 720 simulated transient cycles

� Neural network as objective function

� Minimization of BSFC

� Boundary condition:

NOx,EOP ≤ base variant

� Optimized Parameters:− Lambda Multiplier

− Charge pressure multiplier

− Site density

− SCR catalyst length

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012 | Frankfurt/Main | F. Forsthuber | Slide 20

Optimization – Mean Value Model

Final OptimizationResults

BSFC -2.1%

NOx -4,9%

NOx,EOP -4.0%

Optimized parameter set combined with insulated exhaust pipes

Optimization result verified with detailed model

GT-Suite Users MeetingSimulation of the Tail-Pipe Emissions for a Heavy Duty Diesel Engine in GT-Power

Oct. 22, 2012| Frankfurt/Main | F. Forsthuber | Slide 21

Summary & Outlook

� Summary− Predictive simulation model containing engine and aftertreatment

− Phenomenological Diesel combustion model for fuel consumption and NOx formation

− Aftertreatment with reaction kinetics (NH3 storage and NOx conversion)

− Sensitive to the relevant parameters of engine and aftertreatment operation

− Steady state operating points and transient cycles

− Neural network model for faster runtimes

− Numerical optimization methods

� Outlook− Extended reaction kinetics

− Advanced numerical optimization methods (heuristic algorithms…)

Thank you for your attention !

Dipl.-Ing. Friedrich Forsthuberfriedrich.forsthuber@ifa.tuwien.ac.at