study of after-treatment challenges in hybrid vehicles ... · • partly due to higher density of...

Managed by UT-Battelle for the Department of Energy

Study of after-treatment challenges in hybrid vehicles through system simulations

Zhiming Gao,

Kalyana Chakravarthy,

Stuart Daw

Oak Ridge National Laboratory

Sponsor : Lee Slezak Vehicle Technologies Program

U.S. Department of Energy

2010 DOE-CLEERS workshop April 22, 2010

2 Managed by UT-Battelle for the Department of Energy

Introduction •  Systems simulations

–  Virtual prototyping using component models

–  Component compatibility, integrated controls

–  Inexpensive

•  US DOE Vehicle Systems Analysis Technical Team (VSATT) –  Powetrain System Analysis Toolkit (PSAT) developed at ANL

–  ORNL is tasked with studying after-treatment options

•  Current focus is primarily on hybrid vehicles –  Fuels

–  Operating modes (HCCI, PCCI/HECC, GDI etc.)

–  Alternate aftertreatment options

–  Battery charging strategies


PSAT construct •  Graphical interface written in C# •  Component modules written in Matlab-Simulink, Stateflow •  Model databases managed by XML •  User defined components

Vehicle Configuration

Driving, braking & shifting

Powertrain Components

Simulation Setup and Run

User Define Components

Variable Solver Options

Post Analysis & Simulation Output options

Save & Reload Data

PSAT GUI

XML Model

Database User Front-End

Models


ORNL contribution to VSATT •  Objective : develop engine maps and emissions control device

models for simulating performance of conventional, hybrid and plug-in hybrid vehicles operating with gasoline, diesel and alternative fuels –  Engine maps (steady state) –  Transient engine warm-up model –  Oxycat model –  Lean NOx trap (LNT) model, regen strategies, aging/sulfation effects –  3-way catalyst (TWC) model –  Diesel particulate filter (DPF) model –  SCR model (for Cu-ZSM-5), dosing strategy

•  Approach –  Physically based models to deal with transients (move away from steady state

components of PSAT) –  Generate/utilize public domain lab, engine dynamometer data for building maps

and models –  Fill gaps in experimental data using predictions from analytical, computational

tools such as WAVE, GT-power or in-house software


Engine warm-up model •  Approach : physically based first order

non-equilibrium model (using data from steady state maps) to simulate engine transients

•  Handles cold/warm start (important for hybrid vehicles) for both gasoline/diesel engines

•  Sample results : Cold start UDDS cycle using a Mercedes 1.7L engine (A170 compact car)

•  Published in IJER (Gao et al., Int. J. Engine Res., vol 11, 2010)

Mileage (mpg)

CO (g/mi)

HC (g/mi)

NOx (g/mi)

PM (g/mi)

Experiment 40.3 2.28 0.54 0.74 0.14

Simulation 40.4 2.29 0.54 0.89 0.12


TWC model

•  Model validation conditions –  Data from a gasoline engine from vehicle

tests (supplied by an OEM) –  UDDS cycle with a cold start

•  Integrated emissions : –  CO (g/mi) : 0.833 (exp) vs. 0.836 (model) –  NOx (g/mile) : 0.156 (exp) vs. 0.157

(model) –  HC (g/mile) : 0.139 (exp) vs. 0.148 (model)


SCR model for Cu-ZSM-5 catalyst •  1-D transient simulink module

•  Based on Chalmers/GM data/model

•  NH3 adsorption/desorption

•  3 SCR reactions (NO, NO2, “fast”)

•  NO, NH3 oxidation reactions

•  No N2O (simplicity, same as in LNT model)

•  Cu-ZSM-5 appropriate for low T applications

•  O2 effect, HC inhibition of SCR reactions not included yet

•  Urea thermolysis/hydrolysis also not included yet

•  Predictions in line with published data on a recent commercial formulation

Points from experiments by Olsson et.al, Appl. Cat. B: Environ., 81(2008), 203-217. Lines from ORNL simulation.

SAE 2009-01-0897

Steady State Response

ORNL Model


Gasoline vs. diesel •  SI engines

–  Stoichiometric operation => higher exhaust T –  Most emissions during cold start but faster light-off –  TWC technology still evolving (low PGM, dual cat. Systems) –  May need lean NOx control with GDI

•  Diesel engines –  Very lean operation => low exhaust T –  More efficient than gasoline engines (due to high compression ratios) –  After-treatment more challenging, technologies evolving –  Unconventional modes (HCCI, MK, PCCI/HECC) proposed for emissions/efficiency –  Fuel penalty associated with aftertreatment (LNT, DPF pressure drop/

regenerations)

•  Hybrid vehicle pose additional aftertreatment challenges –  Engine switches on and off several times during a drive cycle –  Small engine size => lower exhaust temperature


Gasoline vs. diesel with out aftertreatment

•  Simulation parameters –  Prius HEV (28% series, 72% parallel) –  Hot start UDDS cycle –  1.3kWhr battery (charge 65%) –  1.5L stoichiometric gasoline engine

based on Miller cycle (map available in PSAT) with TWC

–  1.5L diesel engine (performance scaled down from a 1.7L Mercedes A170 map) with no valving adjustments, no NOx/PM control

•  Results –  84.2 mpg diesel vs. 70.7 mpg gasoline

(SAE 2007-01-0281 reports 71.2 mpg) •  Partly due to higher density of diesel

–  Max efficiency : 41% diesel vs. 37% gasoline

–  Ave. efficiency : 36% diesel vs. 34% gasoline

–  Diesel offer 19% more mpg, 5.4% more energy efficiency (mainly due to compression ratio)


Simulations in present study •  Compare fuel economy, emissions of hybrid and plugin hybrid

electric vehicle (HEV and PHEV) with a similar conventional small car

•  Gasoline engine (with TWC) and diesel engine with LNT and SCR –  Conventional vehicles

•  Honda Civic (1110kg) with a 2.0L engine from SAAB Biopower 9/5 flex-fuel vehicle with 2.7L TWC (gasoline based map used)

•  Mercedes A170 (1090kg) with (its own) 1.7L diesel engine with 2.8L LNT/SCR

–  Hybrid vehicles : Toyota Prius (28% series, 72% parallel) of 1450kg •  Its own 1.5L stoichiometric Atkinson cycle based gasoline engine with a 2.0L TWC •  1.5L diesel engine (scaled down from Mercedes 1.7L engine) with a 2.4L LNT/SCR

–  Battery •  1.3kWhr with 65% initial charge in HEV (charge sustaining mode, final charge almost

similar) •  5.0kWhr with 100% initial charge in PHEV (charge depletion mode), full battery charge

roughly translates to 20 miles (16kWhr battery in 1600kg GM Volt reported to last for 40 miles)

•  Cost of electricity (about $0.1 to $0.15 per kWhr) not factored into mileage comparison

•  Focus on CO, HC in addition to NOx (PM will be added later)


Simulations in present study •  Driving conditions

–  Conventional vehicles and HEVs •  1 cold start UDDS cycle (~1370s, ~8miles) for gasoline vehicle •  5 consecutive UDDS cycles each with cold start (NOx/NH3 storage carry over from a

cycle to next) •  65% initial battery charge (final charge with in 2%-3% range)

–  PHEV •  5 UDDS cycles in succession (engine on rarely during first 3 cycles) •  100% battery charge (5kWhr battery charge roughly translates to 20 miles of driving)

•  Control strategies –  LNT regeneration : optimal of 3 available

•  60s minimum lean phase •  T > 150C (CO poisoning of Pt observed at low T) •  Initiate regeneration when LNT-out NOx exceeds a threshold value •  Rich pulse width adjusted to achieve required NOx conversion

–  SCR urea dosing •  T > 150C (urea conversion to NH3, SCR reactions kinetically limited at lower T) •  Inlet NH3 (from urea) = inlet NOx (from engine map, no NOx sensor)


Results and analysis


Fuel economy

•  Combustion efficiency is about 6% higher in case of diesel –  Gasoline engine in HEV/PHEV is based on an Atkinson cycle –  LNT and DPF add about 2% fuel penalty each –  SCR has no fuel penalty

•  32.5% urea solution costs about same as diesel •  Early studies show 1 gallon of urea solution needed for 18 gallons of fuel

•  GDI modes can increase efficiency of gasoline engines

•  Use of PCCI/HECC modes (when possible) increase efficiency of diesel engines only marginally

conventional HEV PHEV

gasoline 24.5* 67.3 113.3

Diesel with LNT

40.4 (2.1%) 78.8 (2.8%) 133.9 (1.9%)

Diesel with SCR

41.3 80.9 136.4

Numbers in red indicate fuel penalty associated with after-treatment

* Oversized engine optimized for E85


Engine-out emissions •  Engine

–  on 35%-38% of the time in HEVs

–  On 21%-23% of the time in PHEVs

•  Emissions of gasoline engines roughly proportional to fraction of time engine is on

•  HEVs produce more NOx in case diesel engines despite engine being on only a fraction of the time


NOx emissions : conventional vs. HEV •  Engines generally operate in high

efficiency mode in HEVs

•  Gasoline engines operate almost always in stoichiometric mode –  combustion temperature roughly the

same in conventional and HEVs

–  Catalyst temperature little cooler in HEVs (intermittent cool down, lower emissions => lower heating due to reactions)

•  High efficiency operation of diesel engine is associated with high operating temperatures –  Average engine-out temperature,

catalyst temperatures higher for HEVs than conventional engines

–  Higher NOx in case of HEVs

–  PHEVs produce most of the NOx in last 2 cycles, where they are similar to PHEVs

gasoline

diesel


Catalyst-out emissions

•  HEVs powered by gasoline engine lead to more emissions than conventional counterpart (delayed light-off)

•  CO, HC emissions from LNTs very high

•  DOC is needed upstream on SCR to reduce CO, HC (not included in results shown)

•  SCR performance very inadequate (NH3 slip is 0.02g/mile, 0.04g/mile, 0.07g/miles respectively for conventional, HEV and PHEV)


Catalysts performance

•  TWC performance lower in HEVs and PHEVs (delayed light-off)

•  CO, HC conversion in LNTs very low (non-optimal regen)

•  SCR performance low –  Kinetically limited

•  60% (175oC), 80% (200oC) •  Higher in HEVs

–  not NH3 supply limited –  Improvement with upstream DOC

(fast SCR instead of NO SCR) –  May worsen with HC poisoning,

urea conversion are included

•  NOx conversion by LNTs lower in HEVs than in conventional vehicles


Intermittent operation effect on TWC

•  Light-off is delayed by more than 100s in hybrid vehicles (relative to conventional vehicles) –  Run engine at high temperature

(perhaps inefficient mode) till the catalyst lights-off in HEV, excess power can charge the battery


Reductant utilization during LNT regen

•  More than half CO (and HC) emissions from LNT are due to reductant slip during regeneration events –  Partly due to non-optimal regen

strategy •  Increasing rich pulse width beyond a

certain level has diminishing returns on NOx conversion (pre-lightoff NO emissions are not affected)

•  Switch to lean operation when the cat-out flow is rich (UEGO sensor), i.e., non-constant rich pulse width

–  Add a DOC (with stored O2) to reduce CO, HC slip during regen events

CO conversion in LNT on a conventional diesel vehicle


NOx conversion in LNTs : conventional vs LNTs

•  NOx conversion lower in HEVs than in conventional diesel vehicles –  HEVs exhaust hotter on average –  Much of the difference is due to delay in

first regeneration event in case of HEVs –  regen not initiated until Tcat-out > 150oC

(based on observed CO poisoning during rich operation at low T)

•  Regeneration actually efficient at 150oC with H2 but not with CO (with the Umicore GDI CLEERS catalyst)

•  CO and H2 lumped into one species in the current LNT model (may need to be modified)

•  Control strategies are rapid LNT warmup in early stages of engine operation in HEV –  Engine typically operates in high efficiency

(high T) modes in HEVs, increase T further if possible

conventional

HEV


Summary

•  Combustion efficiency of diesel engines is about 6% higher than that of gasoline engines in all vehicles –  LNT add about 2% - 3% fuel penalty (most of it used to reduce stored O2 than

NOx)

–  SCR operational costs not easy estimated •  (high quality 32.5% urea solution roughly same cost as diesel fuel) •  Urea is used only to reduce NOx (no stored O2 as in LNTs) •  Urea to NH3 conversion typically 50% - 70% for light duty vehicles

–  DPF add about 2% fuel penalty (results not shown), higher in HEVs •  Pressure drop effect of power negligible (regenerate when pressure drop reaches

7.5kPa, a strategy resulting in one regeneration every 40 UDDS cycles) •  Engine can not be switched off in HEV/PHEV once regen starts

•  HEVs/PHEVs may need rapid warmup strategies

•  Need focus on low temperature performance of LNT/SCR systems


Future work

•  Use alternate (UEGO sensor based) regen strategy for better reductant utilization in LNTs

•  Add DOC upstream of SCR catalysts –  Oxidize CO, HC, NO

–  Improve SCR performance

–  HC poisoning (long term)

–  Urea thermolysis/hydrolysis (long term)

•  Simulate diesel vehicles with DPF

•  Study effects of increased insulation

study of after-treatment challenges in hybrid vehicles ... · • partly due to higher density of...

Documents