Low-Temperature Gasoline Combustion (LTGC) Engine Research

Low-Temperature Gasoline Combustion

(LTGC) Engine Research – Previously known as HCCI / SCCI –

John E. Dec Jeremie Dernotte and Chunsheng Ji Sandia National Laboratories

June 17, 2014 – 12:00 p.m. U.S. DOE, Office of Vehicle Technologies Annual Merit Review and Peer Evaluation

Program Managers: Gurpreet Singh & Leo Breton Project ID: ACE004

This presentation does not contain any proprietary, confidential, or otherwise restricted information. Overview

Timeline Budget ● Project provides fundamental ● Project funded by DOE/VT: research to support DOE/Industry FY12 – $740k advanced engine projects. FY13 – $720k ● Project directions and continuation are evaluated annually. Partners / Collaborators ● Project Lead: Sandia ⇒ John E. Dec Barriers ● Part of Advanced Engine Combustion working group – 15 industrial partners ● Increase the efficiency of LTGC / HCCI. ● General Motors – in-depth collaboration ● Cummins – spark-plug cylinder heads ● Extend LTGC / HCCI operating range to higher loads. ● LLNL – support kinetic modeling ● Univ. of Calif. Berkeley – CFD modeling ● Improve the understanding of in-cylinder processes. ● Univ. of Melbourne – biofuels & kinetic modeling ● Chevron – advanced fuels for LTGC

Objectives - Relevance

Project objective: to provide the fundamental understanding (science-base) required to overcome the technical barriers to the development of practical LTGC / HCCI engines by industry.

FY14 Objectives ⇒ Increased Efficiency, High Loads, Improved Understanding ● High-load Limits for CR =16: Determine high-load limits for a range of boost pressures (Pin) and speeds. ⇒ Also, potential of Miller cycle to incr. load. ● Noise analysis: Comprehensive study of CNL and RI over range of conds. ● High-Efficiency LTGC: Examine factors affecting measurement of Thermal Efficiency (TE). ⇒ Clarify differences between LTGC and RCCI & CDC. ● Evaluate potential of improving TE over load range with Partial Fuel Stratification (PFS) by optimizing GDI fuel-injection strategy (multi-year task) ● Fuel-Distribution Imaging: Apply PLIF imaging in optical eng. to understand how GDI strategies affect φ-distribution, to help optimization (multi-year task)

● Complete Facility Upgrade for spark-assist, higher GDI Pinjection, & lower swirl ● Support Modeling: Chemical-kinetics at LLNL & CFD at UC-Berkeley & GM. Sandia LTGC Engine Laboratory

● Matching all-metal & optical LTGC research engines. – Single-cylinder conversion from Cummins B-series diesel.

Flame Intake Plenum Arrestor Exhaust Plenum Optical Engine Water & Oil Pumps & All-Metal Engine Heaters Optical Engine Optics Table

All-Metal Dynamometer Engine

● Bore x Stroke = 102 x 120 mm ● 0.98 liters, CR=14 & 16 Unless noted: Ringing ≤ 5 MW/m2 & spd = 1200 rpm NOx & soot emiss. > 10x below US-2010 Approach

● Use a combination of metal- and optical-engine experiments and modeling to build a comprehensive understanding of LTGC / HCCI processes.

● Metal Engine ⇒ high-quality performance data ⇒ well-controlled experiments – High-Load Limits: Adjust CA50 as fueling increased for good stability and no knock. – Noise Study: Analyze combustion noise level (CNL) & ringing intensity (RI) for wide range conditions. Investigate fundamental causes of differences between CNL & RI. – High-Efficiency Studies: 1) Analyze all factors affecting measurement of TE; 2) Sweep parameters to find highest TE; 3) Effects of GDI timing & multiple injections. ● Optical Engine ⇒ detailed investigations of in-cylinder processes. – Fuel Distribution Imaging: 1) PLIF imaging calibrated in-situ; 2) Vertical laser sheet to see all elevations, 3) Obtain φ-map images for various fuel-injection strategies. ⇒ Guide application of PFS in metal-engine for higher TE ⇒ Model validation ● Computational Modeling: 1) Collaborate with UC-B and GM on CFD modeling for improved understanding of PFS ⇒ Results guide experiments for higher TE. 2) Work with LLNL to improve kinetic mechanisms of gasoline/ethanol blends. – Contribution: identify key trends, provide validation data, discussion & feedback. ● Combining techniques provides a better understanding & more-optimal solutions ● Transfer results to industry: 1) physical understanding, 2) improved models. Approach - Milestones

● March 2013  Deduce thermal boundary-layer profiles adjacent to the cylinder-wall from temperature-map images. ● September 2013  Determine the effectiveness of PFS for increasing efficiency above the values obtained with well-premixed fueling for operation with CR = 16:1. ● December 2013  Determine effects of intake temperature, gasoline direct injection (GDI) timing and GDI injection pressure on maximum load and peak efficiency at a representative boosted operating condition. ● March 2014  Establish optical setup and data reduction techniques for fuel-distribution imaging of DI-PFS. ● June 2014 Complete investigation and first-order optimization of single-injection DI-PFS over a wide range of intake boost pressures, including maximum load and peak efficiency at each pressure. ● September 2014 Complete installation and shakedown testing of new low-swirl cylinder head with spark-assist capability. Accomplishments – Overview

● Determined high-load limits for CR = 16 over range of Pin and speeds. – Also, conducted analysis of Miller cycle for its potential to increase load.

● Conducted extensive study of CNL and RI over a range of loads, Pin, CA50, speed & knock intensity. Also, examined & explained reasons for differences. ● In-depth investigation of all potential factors affecting measurement of TE. – Also, conducted additional parameter sweeps to determine effects on TE. – Compared TE of LTGC with RCCI & conventional diesel combustion (CDC). ● Better optimized PFS fuel distributions for improved TE over the load range. – Evaluated effect of GDI timing for a single-injection PFS. – Initial investigation of potential for further improvements with multiple injections. ● Established optical setup and characterized new back-illuminated CCD camera for quantitative fuel-distribution imaging. – On track to obtain φ-map images for initial GDI sweeps this FY as planned. ● Facility upgrade ⇒ on track to complete required modifications to new cylinder head for installation with 300-bar GDI and spark assist this FY. ● Collaborated with UC-B and GM on CFD modeling and LLNL on kinetics. High-Load Limits for CR 16 ⇒ Pin sweep

22 Gasoline,E20, PM PM ● CR = 14 results from last year: E0,E10, PM PM Gasoline,CF-E0, PM PM 20 CF-E0, PM CF-E0,CF-E0, PMDI-60, Tin Tin = 30C= 30C Early-DI PFS gives higher loads with CF-E0, DI-60, Tin = 30C – 18 E0,CF-E0, PM, DI-60, no EGR Tin = 30C CR=16, CF-E0, PM E0,CR=16, PM, CF-E0, no EGR PM [bar] [bar] CR=16, CF-E0, PM less boost (19.4 bar IMEPg at Pin = 3.0) CR=16, CF, DI-60, Tin = 40C g 16 CR=16,CR=16, CF,CF-E0, DI-60, PMDI-60 Tin = 30C CR=16, CF, DI-60, Tin = 30C ● Focus on CR = 16 vs. 14, 14 Gasoline, PM, no EGR ⇒ Using Certification Fuel (CF) 12 10 ● Premixed fueling: CR 16 gives up to 8 Maximum IMEP 13% higher load at low boost vs. CR 14. 6 4 ⇒ Lower req’d. Tin with higher CR. 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0 ● Early-DI-PFS fueling: CR 16 gives higher Intake Pressure [bar] IMEPg for 1.6 < Pin < 2.5 bar. 48 20 18 ⇒ Greater stability allows high φm 46 16 44 14 ● TEs at max. load for each Pin are notably 42

12 [%] higher for CR = 16, both PM, ad DI. 40 10 2 38 8 ● Max. load at high boost is about 9% less 6 36 with CR 16. ⇒ More EGR req’d. to 4 O Exhaust 34 2 control CA50 ⇒ limits air, so lower φm. [%] Thermal Eff. Indicated 32 0 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 ● Still reach 17.7 bar IMEPg at Pin = 3.0 bar Intake Pressure [bar] Effect of Miller Cycle ⇒ CR = 14, ER = 16

Early-DI PFS – 60° CA ● Max. load at high boost (Pin > ~2.5 bar) 22 + ~2.0 %-units less with CR 16. ⇒ More EGR req’d. to 20 (higher TE) control CA50 advance ⇒ limits air & φ . 18

m [bar] 16 g -12.5% ● Would a Miller cycle be better? 14 (less charge-mass) 12 – Reduce req’d EGR & allow higher φm. O2 limit, CR = 16:1 10 – But it would reduce the charge mass. 8 CR = ER = 14, Experimental measurements Maximum IMEP CRCR = = ER ER = = 14, 14, Experimental Experimental measurements measurements 6 CR = ER = 16, Experimental measurements CR = ER = 16, Experimental measurements ● Compute max. IMEP for charge-mass Effective CR = 14 & ER = 16, less charge mass g 4 CR = ER = 16, Experimental measurements EffectiveEffective CR= 14 CR & =ER 14 = & 16, ER less = 16, charge less charge mass &mass higher TE reduction of 12.5% vs. std. valve timing. 2 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 ● Increase load to account for higher TE. Pin [bar]

– TE gain ~2.0 %-units, quite significant. 49 ● Standard cycle gives higher loads except 48 47 for Pin = 3.0 bar. 46 Here, Miller cycle increases max. IMEP – g 45 from 17.7 to 18.1 bar, a marginal gain. 44

CR = ER = 14, Experimental Measurements ThermalEfficiency [%] 43 ● Standard cycle seems better overall. CR = ER = 16, Experimental Measurements Effective CR = 14, ER = 16 42 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 ● Similar results for premixed fueling ⇒ backup slide. Pin [bar] High-Load Limits for CR 16 ⇒ Speed Sweep

● Determine high-load limit for CR = 16, Premixed, Pin = 2.0 bar, Tin = 60°C, Ringing = 5 14 IMEPg,IMEPg, 14:1 14:1 and compare with previous data for IMEPg, 16:1 (no bad point) IMEPg, 16:1 (no bad point) IMEPg, 16:1 (max-Phi-m recorded) IMEPg, 16:1, Pin=2.4 bar (no bad point) CR = 14. ⇒ Premixed, Pin = 2 bar 13 IMEPg, 16:1, Pin=2.4 bar (no bad point) IMEPg,IMEPg, 16:1, 16:1, Pin=2.4 Pin=2.4 b b (max-Phi-m (max-Phi-m recorded)recorded) ● CR = 14: Consistent trend 12 Engine still under control ⇒ Knock/stability limited at all speeds. [bar] (no run-away behavior) g 11

● CR = 16: Trend reverses at 1300 rpm IMEP

– Typical decrease w/ speed, ≥ 1300 rpm. 10 Knock-run- Intemittent "Bad" away/Stabilty Limit Cycle Limit – IMEPg increases with speed ≤ 1300 rpm 9 ⇒ Changes in stability? not understood 800 1000 1200 1400 1600 1800 2000 2200 2400 Engine Speed [rpm] ● ≤ 1300 rpm: Typical knock/stability limit. 5.0 IMEPg,IMEPg, 14:1 14:1 IMEPg,IMEPg, 16:1 16:1 (no (no bad bad point) point) IMEPg, 16:1 (max-Phi-m recorded) 4.5 IMEPg, 16:1, Pin=2.4b (no bad point) IMEPg, 16:1, Pin=2.4b (no bad point) ● > 1300 rpm: Limited by occasional near- IMEPg,IMEPg, 16:1, 16:1, Pin=2.4b Pin=2.4b (max-Phi-m (max-Phi-m recorded) recorded) 4.0 COV-limitCOV-limit = = 2% 2% ⇒ not understood Contains one "bad" cycle misfire cycle, ~1 in 1000 3.5 in the 100-cycle sample. [%] – Remains stable as load increased, but g 3.0 2.5

frequency of near-misfires increases. IMEP - 2.0 Much higher loads could be obtained, – COV 1.5 but risk high COV due to “bad” cycles. 1.0 0.5 ● TE ~1.5 %-units higher with CR = 16 0.0 across speed sweep (not shown). 800 1000 1200 1400 1600 1800 2000 2200 2400 Engine Speed [rpm] Analysis of Combustion Noise Level (CNL)

Log. scale 100 ● Conducted extensive study of CNL and RI Pin = 2.0 bar #4 98 Tin = 30°C ]

2 10 over range of loads, CA50, Pin, & speed. 96 – Knocking and non-knocking conditions 94 #3 5 92 – Complete study in SAE 2014-01-1272. #2 4 90

● Example shows φ sweeps without (left) 3 88 CNL[dB] m RI & with (right) CA50 ctrl to prevent knock. CNL (raw Pcyl.) No EGR 86 2 #1 CNL (filt. Pcyl.) 2 RI ≈ 5 MW/m^2 84 ● RI of 5 – 6 MW/m correlates with onset RI ≈ 5 MW/m2 Ringing Intensity [MW/m CNL (raw Pcyl.) with EGR 82 of knock over wide range of conditions. CNL (filt. Pcyl.) 1 80 ⇒ RI eq’n accounts for effect of Pin, Tin, RPM. 0.30 0.32 0.34 0.36 0.38 0.40 0.42 Charge-mass Equivalence Ratio [φm] ● Here CNL = ~90 dB at knock onset, but 1E+0 -70 #1, Φm = 0.304, RI = 1.9 MW/m^2 #2, Φm = 0.319, RI = 4.9 MW/m^2

different at other conditions (e.g. diff. P ). /Hz] 1E-1 -80

in 2 #3, Φm = 0.412, RI = 5.1 MW/m^2 #4, Φm = 0.332, RI = 9.9 MW/m^2 – Spectral analysis shows CNL dominated 1E-2 Structure attenuation -90 by zero mode ⇒ Press rise from combst. 1E-3 -100

– CNL hardly affected by filter removing 1E-4 -110 st nd Dotted curves: 1 and 2 acoustically resonant modes. 1E-5 low-pass filtered spectra -120 ● RI tracks resonant modes. ⇒ Good for 1E-6 -130

knock control. Poor for overall noise. 1E-7 -140 Strucutre Attenuation [dB]

Power Power Spectral Density [bar "zero" mode (1,0) (2,0) ● CNL ⇒ Good for determining noise from 1E-8 -150 0 2000 4000 6000 8000 10000 combustion event. Not sensitive to knock. Frequency [Hz] Mitigating CNL with CA50 Retard ● CNL dominated by lower frequencies arising from the overall pressure rise with combustion. ⇒ CNL follows magnitude of PPRR (peak pressure-rise rate). ● CNL can be reduced by retarding CA50 to reduce PPRR ⇒ But also affects TE.

● For φm = 0.32, retarding CA50 CR = 14:1 φ φ from 366 to 372 CA reduces m = 0.32 m = 0.42 47.5 98 ] CNL from 96 to 85 dB. Premixed 2 47.0 Tin = 60°C 96 – TE increases until knocking Pin = 2 bar gone, then decreases. ⇒ Only 46.5 TE 94 CNL varies by about 0.6 %-units. 46.0 RI 92

● For φm = 0.42, retarding CA50 45.5 90 CNL[dB] from 373 to 379 CA reduces 45.0 88 CNL from 96 to 85 dB. 44.5 Knock Limit 86 TE falls by 1.4 %-units. RI = 5 MW/m2 – Gross Ind. Therm. Eff. [%] 44.0 84 Ringing intensity[MW/m – Falls monotonically since CA50 364 366 368 370 372 374 376 378 380 effect on TE larger for late CA50. CA50 [°CA] ● Realistically, must keep RI ≤ 5 MW/m2 to prevent knock, ⇒ very irritating sound.

– For φm = 0.32, reducing CNL from 91.5 to 85 reduces TE by 0.6 or 0.7 %-units. – For φm = 0.42, reducing CNL from 90.3 to 85 reduces TE by 0.8 %-units. ● Significant noise (CNL) reduction can be achieved with minimal loss of TE. Factors Affecting Thermal Eff. Measurement

● Previous work investigated how TE 51 Early-DI PFS, Pin = 2.4 bar, Ringing = 5 varies with operating conditions. 50 ⇒ Seek conditions giving highest TE. 49 – Obtained max. TE = 49.2%. 48 47 ● Fueling measurement is also critical. 46 CR 16:1, Tin = 30 C, New I & M #1 CR 16:1,CR Tin 16:1, = 30 Tin C, =New 30 CI & M #1 – Use positive-displacement “Max” meter 45 CR 16:1, Tin = 30 C, New I & M #2 CR 16:1, Tin = 30 C ⇒ CR 16:1,CR Tin 16:1, = 30 Tin C = 40C

Very high precision. [%] Thermal Eff. Indicated CR 16:1, Tin = 40C 44 CR 16:1, Tin = 40C CR 14:1, Tin = 40 C CR 14:1, Tin = 40 C ● Install new flow meter and re-calibrate 43 800 1000 1200 1400 1600

(previous one was 11 years old). IMEPg [kPa] 100 51 ● Discovered that Temp. variation in the lab can significantly affect fuel density 99 50 ⇒ fuel measurement and TE. – Measure fuel temp. and correct. 98 49

– Now data from hot & cool days match. 97 48 CE - without Rho Corr. CE - without Rho Corr. ● Other problems identified and fixed CE - with Rho Corr. 96 TE - without Rho Corr. 47 IndicatedThermal Eff. [%] Combustion Efficiency [%] Efficiency Combustion TE - with Rho Corr. Seals in piston accumulator – fixed TE - no Rho Corr. 130605 – TE - no Rho Corr. 130605 95 46 – Fuel volatility can affect gravimetric calib. 0.24 0.26 0.28 0.3 0.32 0.34 0.36 – Compressibility differences between fuels. Charge mass - Phi [-] Peak Thermal Efficiency

Early-DI PFS, P = 2.4 bar, Ringing = 5 ● Applied all corrections and “fixes”. 51 in 50 ● Re-calibrate using Cert. Fuel, same as 49 experiments (no compressibility issues). 48 – Resolved effect of volatility on calibration 47 46 CRCR 16:1, 16:1, Tin Tin = =30 30 C, C, New New I &I &M M #1 #1 ● Installed a Coriolis meter as a check. CRCR 16:1, 16:1, Tin Tin = =30 30 C, C, New New I &I &M M #2 #2 45 CR 16:1, Tin = 30C, New CF-E0 calib. CR 16:1, Tin = 30 C CR 16:1, Tin = 30 C Indicated Thermal Eff. [%] Indicated Thermal Eff. [%] 44 CR 16:1, Tin = 40C ● Peak T-E = 49.6% for 1200 rpm. CR 16:1, Tin = 40C CRCR 14:1, 14:1, Tin Tin = =40 40 C C Varies only weakly with load (φ ) near 43 – m 800 1000 1200 1400 1600

the peak value. IMEPg [kPa] ● Engine speed has a small effect on the 50.4

max. TE. φm ≈ 0.3 – Slight overall trend of lower max. 50.0 TE with increased speed. – 1300 rpm consistently gives a 49.6 slightly higher TE. 49.2

● Best T-E ⇒ 49.7 – 49.8% Peak Thermal [%] Efficiency 48.8 1000 1200 1400 1600 1800 2000 – CA50 = 366.3°CA, Tpeak = 1511 K Engine Speed [RPM] Optimization of PFS – Single Injection

Early-DI PFS, P = 2.4 bar, RI = 5 ● Operating conditions for high TEs, with 50.0 in single-injection early-DI PFS have 49.5 largely been optimized. However, 49.0

1) Discrepancies found for optimal Tin. 48.5 2) Effect of DI timing not fully explored. 48.0 100%-DI @ 60CAD, Tin = 30°C 47.5 100%-DI @ 60CAD, Tin = 40°C ● Conducted a well-controlled Tin sweep. 100%-DI @ 60CAD, Tin = 50°C CR = 16:1 100%-DI @ 60CAD, Tin = 60°C Certification gasoline #3 Ind. Ind. [%] Thermal Efficiency 47.0 1200 RPM – Vary φ at each Tin to find maximum TE. 100%-DI @ 60CAD, Tin = 70°C m PM, Tin = 60°C Pin = 240 kPa 46.5 ● Very little difference, T = 30 or 40°C 0.22 0.24 0.26 0.28 0.30 0.32 0.34 0.36 in φ ⇒ TE drops consistently for T > 40°C Charge Mass Equivalence Ratio [ m] in TDC BDC IVC 50 100 – TE for premixed lower than DI, same Tin. 49 99 ● Optimal DI timing = 50 or 60° CA. 48 98 – Confirms use of DI-60 (check over range) 47 97 ● Combustion Eff. (CE) trends mirror TE. 46 CR = 16:1 96 Certification gasoline #3 – Explains lower TE for DI-timing ≤ 40° CA. 1200 RPM 45 Pin = 240 kPa 95 Thermal efficiency T = 40°C in Combustion Efficiency [%] – Explains only about half of TE variation Ind. [%] Thermal Efficiency Combustion efficiency Φm = 0.30 for DI-timing ≥ 70° CA. 44 94 0 40 80 120 160 200 240 280 ⇒ Cause of rest not yet understood. Start of Injection [CAD]

Potential of Double-Injection PFS to Increase TE

● Single-injection DI @ 60° CA shows typical trend of TE with φm.

– Repeatability ~ 0.1 %-units for 0.33 ≤ φm ≤ 0.44

injection reduces HRR. 48.5 140123 140124 140127 140206140127 ● Double-injections can 48.0 140311140206 140311140312 140312 (double inj.) better-optimize PFS & 140312 47.5 140313 140313 (double inj.) ⇒ increase TE. 140313 (double inj., match CA50) 140317 140317 (double inj.) 47.0 140317 (double inj., match CA50) 140317 (double inj., match CA50) Thermal [%] Efficiency ● Can TE be increased 140318 140318 (double inj.,inj.) match CA50) further with additional 46.5 140319,140318 (doubleTin = 30C inj., match CA50) 140319, Tin = 30C 140319, Tin = 40C stratification? 140320, Tin = 40C 46.0 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44 0.46 0.48

Charge-Mass Equivalence Ratio [φm] Initial Exploration of Increased Stratification

● Increase late-DI fraction and vary late-DI timing ⇒ use Tin = 40°C for better temperature control. ● Significant improvements possible ⇒ increase TE up to 0.72 %-units over single-injection vs. only 0.25 %-units for dataset on previous slide.

49.0 140319, Single inj. DI@60 ● Amplify scale to better 140320, Single inj. DI@60 Ringing ≈ 5 see trends. 48.8 140319, Double inj. 8% Best Point 48.6 140319, Double inj. 13% ● Second-inj. timings of 140319, Double inj. 20% 310 - 315° CA appear 48.4 140320, Double inj. 20% 140320, Double inj. 25% to be best. 48.2 140320, Double inj. 30% ● Larger DI fraction 48.0 sometimes better, 47.8 but not always. TE for Single-Injection @ 60° CA 140319, Single inj. DI@60 47.6 140320, Single inj. DI@60 – Further study required. 140319, Double inj. 8%

Thermal Efficiency [%] 47.4 140319, Double inj. 13% 140319, Double inj. 20% ● This initial work indicates 140320, Double inj. 20% 47.2 140320, Double inj. 25% ⇒ Double-injections 140320, Double inj. 30% have strong potential. 47.0 29050 100 300150 310200 250 320300 330 ● Further improvements likely. Second-Injection Timing [°CA] Potential of Double-Injection PFS to Increase TE

● Single-injection DI @ 60° CA shows typical trend of TE with φm. – Repeatability ~ 0.1 %-units for 0.33 ≤ φm ≤ 0.44

● Preliminary analysis of φm-distribution images indicates mixture not optimal. ● Double-injection DI using 92.5% @ 60° CA + 7.5% @ 320° CA. ⇒ Increases TE by ~0.2 -0.3 %-units over most of φm range. 50.0 ● Gain is mainly due to Pin = 2.4 bar more advanced CA50 49.5 Ringing ≤ 5 for RI = 5 MW/m2. 49.0 – Improved PFS w/ double injection reduces HRR. 48.5 140123 Improvement with 140124 140127 more optiimal PFS 140206140127 ● Double-injection can 48.0 140311140206 140311140312 140312 (double inj.) better-optimize PFS & 140312 47.5 140313 ⇒ 140313 (double inj.) increase TE. 140313 (double inj., match CA50) 140317 140317 140317 (double inj.) 47.0 140317 (double inj., match CA50)

Thermal [%] Efficiency 140317 (double inj., match CA50) ● Can TE be increased 140318 140318 (double inj.,inj.) match CA50) 140318 (double inj., match CA50) further with additional 46.5 140319,140318 (doubleTin = 30C inj., match CA50) 140319, Tin = 30C 140319, Tin = 40C stratification? 140320, Tin = 40C 46.0 140320 (double inj.) 325 - 30% DI ● Yes! 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44 0.46 0.48 Charge-Mass Equivalence Ratio [φm] Response to Reviewer Comments

1. Reviewers made many positive comments. ⇒ We thank the reviewers. 2. Knocking limit might not be Noise limit. What is the sensitivity of TE to noise reduction below the CNL at the knock limit, by further retarding CA50? – Examined this, and the results on slide 12 show that CNL can be reduced significantly for a fairly modest reduction in TE. ⇒ Sensitivity ∆TE / ∆CNL ≈ 0.8 %-units / 5 dB. – Improved PFS could reduce TE loss for controlling noise. 3. Requested more information on planned intake-port revisions & effect on TE. – Current head designed for SR = 2.3. We use an anti-swirl plate in one port to reduce to SR = 0.9, which might generate additional turbulence and increase heat transfer. – Cummins now makes a head with port geometry directly giving SR = 0.7. These heads are being used for the facility upgrade for spark-assist capability and 300 bar GDI. 4. There were several questions/comments on the need to better understand the relationship of our peak TE to those of other LTC engines & to conv. diesel. – As shown on slides 13 and 14, we have reworked our fuel measurement, and better optimized operating conds. Peak indicated TE is now 49.8 vs. 49.2% reported last year. – These are the highest known efficiencies reported for mostly premixed LTGC. – Peak TE for RCCI & CDC vary with publication. Most recent information from UW shows: H-Duty 54.3%, L-Duty 49% for RCCI and H-D 48.7%, L-D 45% for CDC (no aftertreatment). – Our engine is intermediate size ⇒ peak TE is above L-D RCCI and only ~1 %-unit below an RCCI TE scaled to our engine size. It is well above CDC. ⇒ Will study heat transfer, etc.

– Achieved loads to 19.4 bar IMEPg with gasoline vs. ~15 bar IMEPg for RCCI with gasoline. Collaborations

● Project is conducted in close cooperation with U.S. Industry through the Advanced Engine Combustion (AEC) / HCCI Working Group, under a memorandum of understanding (MOU). – Twelve OEMs, Three energy companies, Six national labs, & Several universities. ● General Motors: Bimonthly internet meetings ⇒ in-depth discussions. – Provide data to GM on boosted LTGC and for modeling PFS-LTGC. ● Cummins, Inc.: Design & fabrication of low-swirl, spark-plug cylinder heads.

● LLNL: Support the development and validation of a chemical-kinetic mechanism for gasoline/ethanol blends, Pitz et al.

● U. of California - Berkeley: Collaborate on CFD modeling of PFS-LTGC.

● U. of Melbourne, Australia: Collaborate on biofuels work & kinetic modeling.

● Chevron: Funds-In project on advanced petroleum-based fuels for LTGC. Future Work Improved PFS-LTGC (multi-year task) ● Continue investigation of multiple injections to better optimize fuel distributions for PFS-LTGC. – Improvement of TE over the load range.

– Potential for extending the high-load limits at various Pin. ● Image fuel distributions in optical engine to guide fuel-injection strategies. ● Guidance from CFD modeling at UC-Berkeley and GM. New Cylinder Head ● Compare TE, load range, heat transfer, etc. ⇒ effect of new low-swirl ports. ● Potential of 300 bar GDI injector for PFS improvements. ● Initial testing of spark-assisted LTGC. Analysis ● Use turbo-charger and friction models from GM to investigate how these real-engine effects would change LTGC performance. Support of LTGC/HCCI Modeling ● Continue to provide data, analysis, and discussions to support: 1) kinetic modeling at LLNL, and 2) CFD modeling at UC-Berkeley and GM. Summary ● Determined the high-load limits for CR =16 for both Early-DI-PFS and premixed fueling over a wide range of Pin and engine speeds. – Max. loads ≥ CR 14 for Pin < 2.5 bar. ⇒ 16 bar IMEPg with Pin of only 2.4 bar. – High req’d EGR limits O2 at high boost. Max. load 17.7 bar vs. 19.4 for CR 14. – Miller cycle shows little potential benefit ⇒ slightly increases load at highest Pin. – Determined load limit for speeds from 1000 – 2200 rpm. ⇒ Discovered that above 1300 rpm, load limited by intermittent partial misfire, a new type of limit.

● Conducted extensive study of CNL and RI over a range of loads, CA50, Pin, speed & knock intensity. ⇒ Showed CNL is not sensitive to knock. – RI tracks resonant modes (knock). ⇒ CNL gives combustion noise, not knock. – Showed CNL reduced significantly by CA50 retard with only a small loss of TE. ● Analyzed and reworked our fuel measurement system and better optimized operating conditions. Peak indicated TE is now 49.8% vs. 49.2% in FY13.

● Well-controlled studies showed the effects of Tin, DI-timing, & speed on TE. ● Initial investigation of double-injection DI-fueling showed that it can significantly improve PFS to increase the TE over the load range. ● Collaborated with chemical-kinetic modelers at LLNL on gasoline/ethanol blends, and with CFD modelers at UC-B & GM ⇒ see Technical B-up Slides Technical Backup Slides LLNL Collaboration – Kinetic Modeling ● In FY13, showed that gasoline/ethanol blends work well for LTGC. ● Showed the importance of intermediate temperature heat release (ITHR) and the increase in ITHR with boost for stable high-load operation. ● Work with LLNL (Pitz et al.) to help them develop and validate kinetic models for gasoline/ethanol blends that work for LTGC and capture trends in ITHR. Experiments Calculations ● P = 100: good 0.01 0.01 in Gasoline #2, Tin = 141 C, CA10 = 368.2 CAD, Phi = 0.38 agreement. 0.008 E20, Tin = 144 C, CA10 = 368.5 CAD, Phi = 0.40 0.008 Ethanol, Tin = 142 C, CA10 = 368.9 CAD, Phi = 0.40

CA] 0.006 CA] 0.006 ● P = 240: ° in 100 kPa 100 kPa captures overall 0.004 0.004 E0 ITHR increase 0.002 0.002 E20 HRR / THRHRR / [1/ HRR / THRHRR / [1/ 0 0 with Pin. E100 -0.002 -0.002 ● Additional work 330 335 340 345 350 355 360 365 370 330 335 340 345 350 355 360 365 370 Crank Angle [°CA] Crank Angle [ CA] 0.01 0.01 needed to E0 #2, Pin = 240 kPa, CA10 = 372.0 CAD, Phi = 0.40 capture small 0.008 E10, Pin = 240 kPa, CA10 = 372.1 CAD, Phi = 0.40 0.008 ITHR E20, Pin = 240 kPa, CA10 = 372.2 CAD, Phi = 0.39 increase CA] CA] 0.006 ° 0.006 differences Ethanol, Pin = 240 kPa, CA10 = 371.76, Phi = 0.425 E0 between E0, 0.004 240 kPa 0.004 240 kPa E10 E10, and E20. 0.002 0.002 E20 HRR / THRHRR / [1/ HRR / THRHRR / [1/ 0 0 E100 -0.002 -0.002 330 335 340 345 350 355 360 365 370 330 335 340 345 350 355 360 365 370 Crank Angle [°CA] Crank Angle [ CA] UC-Berkeley Collaboration – CFD Modeling

● Work with Ben Wolk and J-Y Chen at UC-Berkeley to investigate whether CFD models can capture the effects of PFS? (UC-B funding from DOE-NSF grant.) – Explained PFS and our data showing how it works. – Supplied data and engine geometry models. – Discussion and feedback for improvement. ● Initial results from UC-B capture the reduction in HRR and PPRR with PFS. ⇒ Refinement needed to better match spread of heat release with PFS.

Grid of the Sandia-LTGC Engine for use with CONVERGE™ CFD software Effect of Miller Cycle ⇒ CR = 14, ER = 16

Premixed ● For CR16, more EGR is required to 20 + ∼ 0.8 %-units limit the CA50 advance with boost. 18 (higher TE) ⇒ Limits air, reducing max. load (φ ). 16

m [bar] 14 g -12.5% ● Would a Miller cycle be better? 12 (less charge-mass) 10 – Reduce req’d EGR & allow higher φ . O2 limit, CR = 14:1 m 8 – But it would reduce the charge mass. 6 CR = ER = 14, Experimental measurements Maximum IMEP 4 ⇒ CR = ER = 16, Experimental measurements ● Trade-off Which wins? 2 Effective CR = 14 & ER = 16, less charge mass Effective CR= 14 & ER = 16, less charge mass & higher TE 0 ● Compute max. IMEPg assuming a 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 charge-mass reduction of 12.5% Pin [bar] compared to CR = ER =14. ⇒ Incr. load to account for higher TE. 49 48

● For Pin ≤ 3.0 bar, higher loads with std. 47 cycle, CR = 16 ⇒ Max IMEPg = 15 bar. 46 45 ● For 3.2 ≤ P ≤ 3.4 bar, higher loads in 44 with Miller cycle ⇒ Max IMEP = 15.7 CR = ER = 14, Experimental Measurements g ThermalEfficiency [%] 43 CR = ER = 16, Experimental Measurements bar, a marginal increase. Effective CR = 14, ER = 16, Interpolated - extrapolated 42 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4 ● Standard cycle seems better overall. Pin [bar]