Annual Report 2006 for Centre of Competence Combustion Processes, KCFP at Lund University

Citation preview

Annual Report 2006 for

Centre of Competence Combustion Processes, KCFP at Lund University

Director: Bengt Johansson [email protected]

2

Table of content KCFP general information _____________________________________________5 Background/history ______________________________________________________ 5 Goal ___________________________________________________________________ 6 Fields of research ________________________________________________________ 6 Organization ____________________________________________________________ 7 Budget _________________________________________________________________ 9

Partially premixed combustion, PPC ____________________________________11 Personnel______________________________________________________________ 11 Publications____________________________________________________________ 11 The effect combustion chamber geometry on HCCI combustion rate ____________ 11 Chemiluminescence imaging of the disc Geometry___________________________________ 13 Chemiluminescence imaging of the bowl Geometry __________________________________ 13

Future Work___________________________________________________________ 14 References _____________________________________________________________ 14

Spark Assisted Compression Ignition SACI _______________________________16 Personnel______________________________________________________________ 16 Background____________________________________________________________ 16 Experimental Apparatus _________________________________________________ 16 Results ________________________________________________________________ 17 The Early Flame Development in Spark Assisted HCCI Combustion_____________________ 17 The effect of swirl on SACI_____________________________________________________ 19

Work in progress _______________________________________________________ 20 High speed Fuel LIF of SACI combustion__________________________________________ 20

Published results _______________________________________________________ 21 Results to be published __________________________________________________ 21

HCCI Combustion Control ____________________________________________22 Persons involved ________________________________________________________ 22 Publications____________________________________________________________ 22 Experimental setup _____________________________________________________ 22 Studies performed ______________________________________________________ 24 Mapping and operating strategy with NG / n-heptane and VGT turbo ____________________ 24 Virtual Heat Release Sensor with FPGA Technology _________________________________ 25 Self-Tuning Cylinder Pressure Based Heat Release __________________________________ 27

Modeling of HCCI Combustion ________________________________________29 Personnel______________________________________________________________ 29 Background____________________________________________________________ 29 Investigation of the temperature stratification and in-cylinder turbulence ________ 29 Effect of Turbulence on HCCI Combustion _________________________________ 31 Future Work___________________________________________________________ 34

3 Publications____________________________________________________________ 34

SI Gas Engine Project ________________________________________________35 Persons involved ________________________________________________________ 35 Publications____________________________________________________________ 35 Experimental setup _____________________________________________________ 35 Studies performed ______________________________________________________ 36 Planned activities _______________________________________________________ 36

KCFP subproject: Fuel Effects _________________________________________37 Personnel______________________________________________________________ 37 Introduction ___________________________________________________________ 37 Fuel Specific Differences between Euro Diesel Fuel and RME under partially premixed combustion____________________________________________________ 37 Fuel Characteristics ___________________________________________________________ 38 Sauter Mean Diameter_______________________________________________________ 38 Distillation Curve __________________________________________________________ 38 Results _____________________________________________________________________ 39 Influence of EGR and Fuel Type ______________________________________________ 39 Influence of Rail Pressure ____________________________________________________ 41 Conclusions _______________________________________________________________ 42

Future work ___________________________________________________________ 43 References _____________________________________________________________ 43

The Generic Diesel Research (GenDies) project ___________________________45 1.

Introduction _______________________________________________________ 45

• Diesel engine flow measurements, Combined measurements of flow field, partially oxidized fuel and soot and, and temperature and heat flux measurements 45 •

• High-speed video campaign for measurements of simultaneous OH and natural luminescence and temperatures measurements with two-color pyrometry _________________ 46

Comparison between optical and all metal engines _______________________ 46

2. Gendies staff _________________________________________________________ 46 3. Diesel engine flow measurements, Combined measurements of flow field, partially oxidized fuel and soot and, and Temperature and heat flux measurements 46 3.1 Diesel engine flow measurements _____________________________________________ 46 Conclusions from Diesel engine flow measurements _______________________________ 50 3.2 Combined measurements of flow field, partially oxidized fuel and soot ________________ 50 Conclusions from combined measurements of flow field, partially oxidized fuel and soot __ 52 3.3 Temperature and heat flux measurements _______________________________________ 53 3.4 Future prospects, outlook for 2007 and beyond ___________________________________ 54

List of publications produced and/or presented during 2006 ___________________ 54 PhD thesis __________________________________________________________________ 54 MSc thesis / Diplomarbeit ______________________________________________________ 54 Papers______________________________________________________________________ 54

4. Optical Analysis of Soot Reduction by Post Injection under Dilute Low Temperature Diesel Combustion __________________________________________ 55 4.1 Introduction ______________________________________________________________ 55 4.2 Experimental setup_________________________________________________________ 56 4.3 Methodology _____________________________________________________________ 59 4.4 Results and discussion ______________________________________________________ 60

4 4.5 Conclusions ______________________________________________________________ 63 4.6 Reference ________________________________________________________________ 63

5. Study of Flame Lift-Off Length in Relation with Engine-out Soot in Heavy Duty Diesel Engine. __________________________________________________________ 64 6. Comparison between optical and all metal engines _________________________ 67 6.1 Personnel ________________________________________________________________ 67 6.2 Background ______________________________________________________________ 67 6.3 Known differences _________________________________________________________ 67 6.4 Test Conditions ___________________________________________________________ 68 6.5 Results __________________________________________________________________ 68 6.6 Ongoing and future work ____________________________________________________ 70

7. High-speed video campaign for measurements of simultaneous OH and natural luminescence and temperatures measurements with Two-color pyrometry. ______ 70 Staff _______________________________________________________________________ 70 7.1 Introduction ______________________________________________________________ 70 7.2 OH luminescence and simultaneous OH and natural luminescence using a stereoscope____ 71 7.3 Two-color pyrometry _______________________________________________________ 73

5

KCFP general information Background/history The centre of competence for combustion processes, KCFP, started July 1 1995. The overall goal of the centre is to develop new modelling and diagnostics tools for better understanding of combustion processes. Activities for intermittent combustion like in an IC engine and continuous combustion like in gas turbines, burnes etc. were included. The first phase of the ATAC (HCCI) project within phase 2 of the competence center started July 1 1997 and ended July 1 2000. In that phase, two Ph.D. students where working full time within the project and optical diagnostics equivalent to 25% of a student were conducted. Per Amnéus was focusing on chemical kinetics and Magnus Christensen on engine experiments. Volvo Trucks, Volvo Cars, Scania, Wärtsilä Diesel and the Swedish Gas Center, SGC sponsored the project as well as the Swedish government. The second phase of the project within phase 3 of the competence centre started July 1 2000 and ended July 1 2003. In this phase three students have been working full time on engine experiments, combustion control and chemical kinetics. A half time student on laser diagnostics was also been involved in the project. Volvo Trucks, Volvo Penta, Volvo Cars, Scania, Saab, Caterpillar, Cummins and Wärtsilä Diesel sponsored the project as well as the Swedish government from the beginning. Later Toyota and Hino Motors joined the project with additional funding. In phase 4 of the competence centre 2003-2005, four full time Ph.D. students could be active in the project, one each for the four fields of research: engine tests, optical diagnostics, kinetic modelling, and combustion control. During the phase 4 Nissan also joined as additional member of the project. From start 1997 the project generated a number of research results being world first. The HCCI research in Lund is now established as leading in the open research community.

Figure 1: Papers on HCCI 1999-2001 (left) and 2001-2004 (right) according to IFP. The Gendies project started in 2001 as part of the CECOST national framework for combustion research. This project has been sponsored by Volvo Cars, Volvo Powertrain and Scania from the beginning. The project focused on generic knowledge

6 of diesel combustion from the beginning but has evolved into studying combustion processes with low emission of NOx. Thus the ATAC (HCCI) and gendies projects have evolved to become similar and it is thus now proposed to join forces in phase 5 of the competence centre.

Goal The main goal of this centre is to better understand the combustion process in internal combustion engines. Of particular interest are the combustion processes with low enough temperature to suppress formation of NOx and particulates, PM, often called Low Temperature Combustion, LTC or homogeneous Charge Compression Ignition, HCCI. To improve our knowledge, advanced measurement systems as well as models will be applied.

Fields of research The proposed centre of competence will focus on the combustion process in internal combustion engines. The overall goal for this is to become a world leading centre of expertise in the current and future combustion processes. A good knowledge of the traditional premixed spark ignited combustion process in gasoline engines will be maintained but we do not see a major effort in investigating this process further but will put some effort in the control of highly diluted SI combustion. Instead the combustion processes which can suppress the formation of nitric oxides by reducing the maximum temperature will be of major interest. Those are sometimes called Low Temperature Combustion, LTC but also the acronym HCCI is used to describe the same type of processes even though the charge can be far from homogeneous in some cases. table 1: Matrix of combustion processes and types of activities. Spark Assisted Compression Ignition, SACI

Partially Premixed Combustion, PPC

Generic Diesel Combustion, GenDies

Highly diluted Spark Ignition (Gaseous Fuel)

Engine tests Laser diagnostics Modelling Combustion control Fuel effects Four major types combustion will be studied, Spark assisted compression ignition, SACI and partially premixed combustion, PPC. Also “traditional” diesel combustion will be studied in the generic diesel program. Finally the combustion process with highly diluted spark ignition will be of interest for gas engines. Five types of activities will be included. The four from the previous HCCI project and a new fifth. Basic engine studies for investigating emissions and efficiency for a range of operating conditions. Laser diagnostics will be used to investigate the processes in more detail. Modelling of the in-cylinder processes will be done with models based on chemical kinetics and fluid dynamics. Active control of the combustion process will

7 be continued with more detailed dynamic models and faster signal processing. The new fifth activity deals with fuel effects i.e. the interaction between fuel and combustion process.

Organization The centre of competence will be organized in a similar way as before with some small changes. A director will organize the work and report to a board consisting of members from participating companies, STEM and academia. Each activity within the centre will have a working group that will meet at least twice annually. At these meetings the most recent results will be presented and also suggested future work will be discussed. A project manager will be responsible for each activity and reports to the director and board. The board will meet as today three to four time per year. The board consist of the following persons: Sören Udd, Volvo Powertrain as chairman with Carina Johansson as suppliant. Börje Grandin, Volvo Cars with Lisa Jacobsson as suppliant Tommy Björkqvist GM powertrain Sweden with Annika Kristoffersson as suppleant Urban Johansson, Scania CV with Astrid Simovic as suppleant Gunilla Jönson, LTH Marcus Aldén, LTH Per Tunestål, LTH Bernt Gustavsson, LTH This sum up to 7 male and 5 female members of the board, making the female participation 41.6%. During the build-up of the Gendies project within Cecost the budget was limited. This meant that during phase 1 of this project we only had funding for part time activities and the rest of the time the test cell was available for the industry to do own experiments. This turned out to be very fruitful as personnel from industry took active part in work in the engine test cell. During the second phase of Gendies this type of activity was included in the budget. The time available for each company was proportional to the amount of funding. We propose to use the same type of arrangement for phase 5 of KCFP. A two month period per million of SEK funding is proposed. During this time period, a test cell is available with the technicians at hand for helping out with the experiments. The cell can be closed and results do not need to be reported outside the company. This means that hardware closer to production could be tested with the state of the art techniques available in Lund. A centre administrator will help and assist the director and be responsible for the administration of the centre.

8 Board Sören Udd, Volvo Powertrain (Chair) Gunilla Jönsson, LTH Börje Grandin, Volvo Cars Marcus Aldén, LTH Tommy Björkqvist, GM Sweden Per Tunestål, LTH Urban Johansson, Scania CV Bernt Gustavsson, STEM

Director Bengt Johansson

Administrator: Maj-Lis Roos

Engine experiments Bengt Johansson / Anders Hultqvist

Laser diagnostics Anders Hultqvist / Mattias Richter

Modelling Xue-Song Bai

Controls Per Tunestål

Fuels Rolf Egnell

SACI exp ½ Håkan Persson

SACI optical engine stud. ½ Håkan Persson

LES Modelling student R. Yu

FPGA control student Carl Willhelmsson

PPC exp. student ½ Andreas Vressner

PPC optical engine student ½ Andreas Vressner

Kinetics student Martin Tunér

Control modelling ½ Control student

SI gas engine student Mehrzad Kaidaki

SACI laser diagnostics student ½ Johan

PPC Laser diagnostics student Johannes Lindén

Gendies optical engine student 1 Leif Hildingsson

Gendies optical engine student 2 Ulf Aronsson

Gendies laser diagnostics student 1 Johan Sjöholm

Gendies laser diagnostics student 2 ½ Johan

Fuels student Uwe Horn

9

Budget Funding of KCFP will be in two parts: a) the stipulated 3 times 7 MSEK per year and b) an addition for the Swedish gas centre part of the activities. The annual income is listed in the table below first without SGC then the SGC part and finally the total budget framework. Funding kSEK Volvo Cars Volvo Powertrain GM/Saab Scania Toyota Nissan Caterpillar Volvo Penta Cargine Finnveden Mecel Loge Wärtsilä Chevron Gas companies SUM Industry STEM SGC LTH TOTAL

Cash 550

Inkind 450

500 550 640 550 550 468 100

322 600 600

300 300 300

sum wo SGC 1 000

SGC cash

822 1 150 1 240 550 550 468 100 300 300 300

Inkind SGC

378

378

120

120

4 858 7 000

7 430

7 000 21 430

700 600 720

300 800

800

300 300

1 298

300 1 598

300 5 158

100 1 398

743 100 2 441

7 000 743 1 500 14 401

7 000 5 500 8 072

500 550 640 550 550 468 100

100

743 1 500 13 358

Inkind total 450

300

100 550

2 572

cash total 550

300 800

100 550

sum SGC

1 043

Total 1 000 1 200 1 150 1 360 550 550 468 100 300 300 300 800 100

3 870

300 9 028

5 600 9 470

7 000 743 7 100 23 871

10 The annual expenses are given below. The budget is showing the costs for common resources in the form of senior researchers. Each unit of 750 KSEK corresponds too 75% of full time for an Asc. Professor. The five fields of research and the four combustion processes are presented and the costs of Ph.D. student salary, technician salary and material are presented for each. The cash budget is indicated in column 7. In column 8 the number of persons active is given. Then the in kind contribution from industry is presented in column 9 followed by the inkind contribution from Lund University. Column 10 is a comment to column 9 indication the type of contribution. Annual budget kSEK incl. OH Engine expr. Phd. Student Technician Material Travel Laser diagn. Phd. Student En Phd. Student FF Technician Laser mtr. Material Travel FF Travel Model Phd. Student Technician Material Travel Control Phd. Student Ph.D. student control Technician Material Travel Fuels Phd. Student Technician Material Travel Adm 5% Sum

Com Ass. Prof. 750

SACI

325

PPC

Gen dies

650 325 75 20

325 325 75 10

100

325 650 325 100

10 10

10 20

650 975 325 100 100 30 20

75 10

SI gas

650 325 75 20

750 325 325

375 650 100 20 750 325

325

325 163 75 20

163 75 10

325

325

70 10

70 10

2 493

4 248

375

656 3 656

2 935

1 070

cash SUM

# of pers

750 1 950 975 300 40 750 1 300 1 950 650 300 100 50 70 375 650 0 100 20 750 650

0,75 3,00 1,50

325 325 150 30 375 650 0 140 20 656 14 401

0,50 0,50

Ind inkind

LTH inkind

950

800 300

120

100

0,75 2,00 3,00 1,00

200 600

0,38 1,00

600 800

0,75 1,00

Prof. A.Prof.

1 600 300

Prof. A.Prof.

800 300

Prof. A.Prof.

800 300

Prof. A.Prof.

300

Prof. A.Prof.

5 600

23 871

600 0,38 1,00 0,00

17,50

3 870

11

Partially premixed combustion, PPC Personnel This work was conducted by Ph.D. student Andreas Vressner under supervision of Professor Bengt Johansson, Ass. Prof. Anders Hultqvist and Rolf Egnell.

Publications Papers: “Visualization of Laser Assisted HCCI Combustion with Natural Gas as Fuel” Fisita/JSAE Technical Paper F2006P206 By Andreas Vressner, Anders Hultqvist, Bengt Johansson, Martin Weinrotter, Ernst Wintner, Kurt Iskra and Theo Neger. Presented by Andreas Vressner at the FISITA World Automotive Congress, Yokohama, Japan, October 2006 “Study on Combustion Chamber Geometry Effects in an HCCI Engine Using High-Speed Cycle-Resolved Chemiluminescence Imaging” SAE Technical Paper 2007-01-0217 By Andreas Vressner, Anders Hultqvist, Bengt Johansson To be presented by Andreas Vressner at the SAE World Congress, Detroit, USA in April 2007 Licentiate Thesis: “Homogeneous Charge Compression Ignition (HCCI) – Combustion Initiation and Diagnostics” By Andreas Vressner Presented November 16th 2006 in Lund, Sweden ISRN LUTMDN/TMHP –06/704--SE

The effect combustion chamber geometry on HCCI combustion rate This a continuation of the work of Christensen et al. within KCFP where different combustion chamber geometries where used to decrease the rate of heat release and thus pressure rise rates in HCCI combustion [2, 3]. Since the whole charge is oxidated gradually the combustion is usually covered in less than 10 CAD and the pressure rise rate therefore high. The pressure rise rate is the limiting factor for maximum load in HCCI engines due to excessive noise or hardware damage. By using a deep square bowl in piston geometry the combustion duration was twice as long compared to a disc combustion chamber. This resulted in half the rate of heat release and half the pressure rise rate. The maximum load of the engine could thus be increased. The reason for this behavior was not entirely known but suspicions were increased heat transfer due to turbulence in combination with the larger wall area of the bowl. Another suspicion was trapped residuals in the bottom of the bowl igniting the charge and propagating throughout the rest of the combustion chamber. Since it was unclear how and where the combustion propagates further studies were needed. The studies performed this year have been fuel tracer Lacer Induced Flourescence

12 (LIF), chemiluminescence imaging, and Large Eddy Simulations (LES). This part of the report covers the chemiluminescence experiments.

Figure 2: Drawings showing the cross sections of the square bowl in piston to the left and the disc shaped piston to the right. The light grey areas are made of quartz glass while the dark grey areas are made of titanium. The Scania D12 engine with optical access where equipped with both combustion chamber types as seen in Figure 2. Both piston crowns are made of quartz glass and optical access is enabled both from beneath and from the side in the bowl case. Very strong glue is the only thing preventing the crown to come off during operation. Both combustion chambers have the same compression ratio (17.2:1) but different wall areas and heat conductivity. Different operation points where selected and evaluated with both geometries to be able to compare the combustion process in the two combustion chambers. In Figure 3 the pressure and rate of heat release (ROHR) histories can be found for a case of 2.5 bar IMEP, and CA50 of 8 CAD ATDC. As seen the combustion duration is much longer in the bowl case. The ROHR and pressure rise rate is about half in the bowl case. The inlet temperature was about 20 °C lower in the bowl case suggesting temperature stratification throughout the bowl where the locations with highest temperature ignite first and propagates to the cooler ones. The resulting combustion duration is the much longer compared to a less stratified charge where all ignites at once. 50 Disc Bowl 200

Disc 30

Bowl

150

20 100 10 0 -40

ROHR [J/CAD]

Pressure [Bar]

40

50 -20

0 CAD ATDC

20

0 40

Figure 3: Pressure and rate of heat release histories. 2.5 Bar IMEP, λ was 3.3 and the combustion phasing in terms of CA50 was 8 CAD ATDC for both geometries.

13

Chemiluminescence imaging of the disc Geometry

TDC

+3

+4

+5

+6

+7

+8

+9

+10

+11

+12

+13

+14

+15

+16

Figure 4: Single cycle chemiluminescence images for the disc case. λ = 3.3, CA50 = 8 CAD ATDC, 2.5 Bar IMEP and an engine speed of 1200 rpm. In Figure 4 chemiluminescence images of the combustion in the disc case can be found. As seen, the combustion first appears in the left part of combustion chamber propagating throughout the charge due to numerous new ignition kernels, typical for HCCI combustion. The combustion duration is about 10 CAD.

Chemiluminescence imaging of the bowl Geometry In Figure 5 chemiluminescence images of the combustion in the bowl case can be found. This is the same load case and combustion phasing as in the disc case previously. In the bowl case the combustion starts much earlier, e.g. 12 CAD BTDC. The combustion from ignition to TDC behaves in a stratified manner almost flame front like even no external ignition source has been used. This suggest stratification in temperature where the zone in the upper right corner of the charge is hottest and where it ignites first. There few new ignition kernels compared to disc case. As the piston passes TDC the combustion speeds up and the rest of charge in the bowl is consumed more rapidly. Lastly some 10 CAD ATDC combustion is seen in the squish volume. The late appearance of intensity in the squish is probably due to quenching because of the large area to volume ratio with resulting high heat losses. As the piston moves down in the expansion stroke the conditions improve and combustion occurs here too. The reason for the faster combustion inside the bowl after TDC is probably due to a less stratified charge at this point. The more uniform charge after TDC might be due to improved mixing by the squish motion and turbulence appearing in this time window. So to summarize three combustion stages have been discovered: 1. Slow and stratified combustion between 10 CAD BTDC and TDC

14 2. 3.

Fast more normal HCCI combustion gradually oxidating the charge in the bowl from TDC to 10 CAD ATDC. Late combustion in the squish probably due to quenching

-14

-12

-10

-8

-6

-4

-2

TDC

+2

+4

+6

+8

+10

+12

+14

+16

+18

+20

Figure 5: Single cycle chemiluminescence images for the bowl case. λ = 3.3, CA50 = 8 CAD ATDC, 2.5 Bar IMEP and an engine speed of 1200 rpm. For more reading about these experiments and a more comprehensive discussion on the combustion behaviour in the bowl the reader is referred to [4, 5]

Future Work The future work of this project involves performance test in the Scania D12 engine in metal configuration. Different combustion chamber geometries will be evaluated regarding maximum load, efficiency, emissions etcetera. The combustion chambers evaluated will be; disch, square bowl, circular bowl and a standard diesel engine bowl geometry. The next step is to convert the engine back to direct injection but this time with common rail to be able to run partially premixed combustion (PPC) in combination with variable valve actuation (VVA). Our hope is to be able to suppress NOx and soot emissions at the same time in this intermediate concept between Diesel and HCCI combustion.

References 1.

Andreas Vressner, Anders Hultqvist, Bengt Johansson, Martin Weinrotter, Ernst Wintner, Kurt Iskra and Theo Neger: “Visualization of Laser Assisted HCCI Combustion with Natural Gas as Fuel”, Fisita/JSAE F2006P206

2.

Magnus Christensen, Bengt Johansson and Anders Hultqvist, “The Effect of Combustion Chamber Geometry on HCCI Operation”, SAE Technical Paper 2002-01-0425

3.

Magnus Christensen and Bengt Johansson, “The Effect of In-Cylinder Flow and Turbulence on HCCI Operation”, SAE Technical Paper 2002-01-2864

15 4.

Andreas Vressner, Anders Hultqvist, Per Tunestål, Bengt Johansson, “Study on Combustion Chamber Geometry Effects in an HCCI Engine Using High-Speed Cycle-Resolved Chemiluminescence Imaging”, SAE 2007-01-0217

5.

Andreas Vressner, “Homogeneous Charge Compression Ignition (HCCI) – Combustion Initiation and Diagnostics”, Thesis for the degree of Licentiate in Engineering, Lund University, 2006

16

Spark Assisted Compression Ignition SACI Personnel This work was conducted by Ph.D. student Håkan Persson under supervision of Professor Bengt Johansson.

Background Spark Ignition (SI) engines struggle with low efficiency at part load operation. A feasible way to improve low load efficiency for cars currently running with SI engines would be to use an engine that can run in HCCI mode at part load and switch to SI at high load. Studies have reported high efficiency and low NOx emissions compared to the SI-engine. To achieve mode switching from SI to HCCI and vice versa more knowledge is needed about the intermediate region where both combustion modes coexists e.g. SACI combustion which is the main target for this project. Also the usage of spark assistance is of interest to control combustion timing. Finally the usage of spark assistance makes it possible to increase the possible operating range in HCCI mode. This work is a continuation of the “Un-orthodox Otto” project that was also dealing with gasoline HCCI, but then mainly in a multi cylinder engine with pressure analysis as the most important tool for understanding and interpret the combustion event.

Experimental Apparatus The experimental engine is based on a Volvo D5, which is a passenger car size five cylinder compression ignition (CI) engine. The engine is converted to single cylinder operation with a Bowditch piston extension. The remaining four pistons are motored and are equipped with counter weights to compensate for the extra weight of the piston elongation on the operated piston. A 58 mm diameter quartz glass is fitted in the piston extension resulting in a flat piston crown. The optical access is approximately 51 % of the total combustion chamber area. Also small vertical windows are mounted perpendicular in the pent-roof giving see-thru access to the vicinity of the sparkplug. The combustion chamber is of SI type with a high pent-roof, four valve design. The Engine is fitted with a pneumatic valve train system supplied by Cargine Engineering. The system features fully flexible valve lift, duration and timing as well as deactivation individual for each valve. The engine with the valve actuators mounted on the cylinder head is shown in Figure 6. Due to optical constraints the engine is run with a slightly lower than normal compression ratio (CR = 9:1) and a cooling water temperature of 70° C. Therefore a fuel blend with lower octane number than normal has to be used. A blend of 40 % ethanol and 60 % n-heptane (mass percentage) is used. The choice of ethanol instead of iso-octane is due to that the former seems to cancel the low temperature reactions (LTR) normally associated with n-heptane.

17

Figure 6. Single cylinder engine with optical access and pneumatic free valve train.

Results The Early Flame Development in Spark Assisted HCCI Combustion The effect of spark assistance is investigated by simultaneous pressure and high speed chemiluminescence imaging. The behaviour of the combustion initiation is observed by changing three different main parameters. 1. NVO. By going from low NVO (0 CAD) to high (200 CAD) the initial charge temperature is increased. As a side effect the fresh charge of fuel and air is diluted when the NVO is increased. 2. Spark timing. The spark timing is advanced when operating at high NVO and low load. 3. Load. A load sweep is conducted with SACI combustion at high NVO, affecting both the residual fractions and temperature. Close to and also past the normal dilution limit for SI combustion in terms of residual dilution, it is possible to achieve an effect from the spark and thereby run the engine in SACI mode. One attempt to explain the effect is that the locally raised temperature from the spark kernel forms a hot spot that provokes the charge to auto ignite. By looking at a pressure trace from SACI operation it is obvious that there is more to it. Characteristic is an initial slower heat release related to the spark. Later the well known fast heat release related to HCCI combustion takes over. This suggests an initial SI flame propagation and heat release with an elevated temperature that triggers auto-ignition and the subsequent combustion in HCCI mode. This is supported by the chemiluminescence imaging conducted for both low and high residual rates. The calculated flame expansion speed for increasing negative valve overlap thus increased residual rate is shown in Figure 7. A decreased in speed, but prevailing flame expansion is seen also for high residual rates.

18

Flame expansion speed [m/s]

12 10 8 6 4

NVO 040 NVO 080 NVO 120 NVO 160 NVO 200

2

0 −18 −16 −14 −12 −10 −8 −6 −4 Crank angle [CAD ATDC]

−2

Figure 7. Mean flame expansion speed for increasing NVO, average of twenty cycles.

13

14 CA50 COVIMEP

12

CA auto ign. 11

10

9

8

7

6

5

4

3

2

1

−60

−55

−50 −45 −40 −35 Spark angle [CAD ATDC]

−30

COVIMEP [%]

CA50, CA auto ign. [CAD ATDC]

15

0

Figure 8. CA50, calculated auto-ignition timing and COV as a function of advanced spark timing. SACI combustion at these high residual rates past the lean limit of the conventional SI engine is thought to be enabled by the elevated temperature conditions. The early flame development is a balance between the heat released by the flame and the convective heat losses to the surrounding gas. The residuals that dilute the charge will lower the heating value per unit mass, lowering the flame temperature. However, with trapped residuals the initial temperature is higher, thus the effect of heat transfer should be lower. Also a lower flame speed can be tolerated in SACI compared to in the SI case. For the SI engine, flame speed has to be kept high enough to prevent burn

19 duration from increasing. Too long burn duration means at first, lower efficiency and, in severe cases risk of partial misfire. In SACI the early flame is used only to initiate the greater part of the charge to auto ignite and burn in HCCI mode, therefore slower flame propagation can be accepted in SACI mode. As can be seen in Figure 8 it is possible to phase combustion also at as low load as 2.5 bar IMEPnet when highly diluted with residuals. It should be noted that both combustion timing (CA50) as well as auto ignition timing is affected.

The effect of swirl on SACI The high influence of turbulence on SI combustion is well known. Increasing turbulence levels wrinkles the growing flame, increasing the flame surface, thus the expanding flame area. Further it is well documented that the turbulence levels increases with inlet port deactivation in the used combustion chamber geometry. The question that remains is if turbulence plays the same role for residual diluted SACI combustion. For NVO HCCI and SACI combustion the state of the residuals in the current cycle is highly depending on the previous cycle when it comes to both composition and temperature. This could indeed influence the conditions for the growing flame as it has been seen to give raise to oscillation phenomena in HCCI combustion timing Figure 9 shows the correlation between the position of CA 1 % heat released and turbulence level at the spark plug position. When turbulence is increasing the heat release is advanced. The correlation coefficient at this position is 0.39. This behaviour is consistent if looking at both different CADs and measurement points. At again higher turbulence levels the effect is again the opposite and the flame growth is delayed. This effect is however not as pronounced for all measurement points. The again retarded flame growth for even high turbulence levels could be due to increased heat losses when the early flame is to heavily wrinkled with high area to volume ratio. It can be concluded that the turbulence level still plays an important role for the early flame propagation in SACI combustion at high residual rates although cyclic influences from the residuals. 0

CA at 1 % Heatrelease

−1 −2 −3 −4 −5 −6 −7 −8 0

1

2 Turbulence [m/s]

3

4

Figure 9. Correlation between CA at 1 % heat released and turbulence at -20 CAD ATDC, corr. coeff = 0.39.

20

Flame Expansion Speed [m/s]

10

8

Ref. case 1 Intake 1 off Intake 2 off Ref. case 2

6

4

2

0 −16

−14

−12

−10 −8 CAD [ATDC]

−6

−4

−2

Figure 10. 202 CAD NVO, stoichiometric conditions. Reference case with both inlet valves activated. Increased turbulence by inlet valve deactivation has been shown to influence SACI combustion in both of the two combustion modes. The interpretation of the results from the initial SI combustion is relatively straight forward. As can be seen in Figure 10 the flame expansion speed can be increased almost to the levels of normal SI operation without residual dilution. To fully understand the influence of the auto ignition process requires more knowledge of the residual stratification and its response to the mixing process by valve deactivation. However it is clear that auto ignition is delayed and the HCCI reaction rate is slowed down. One suggestion to the retarded combustion timing at high residual rates could be that the stratification between the hot inert residuals and the fresh stoichiometric charge is lowered by the increased mixing of the swirling motion thereby lowering the reactivity of the total charge. Auto ignition can be expected to occur in the border between the hot residuals and the fresh charge. Increased mixing with the fresh charge will lower the temperature, delaying auto ignition.

Work in progress High speed Fuel LIF of SACI combustion To gain more knowledge of the effect of stratification on SACI combustion the effect of fuel and temperature stratification is studied. Stratification is achieved by moderating temperature and fuelling in the two intake ducts as well as with direct injection. Figure 11 shows some initial results of laser induced florescence (LIF) in SACI combustion. Ethanol is used as fuel with 10 % acetone as fuel tracer. The method gives information of both the flame growth as well as the auto ignition behaviour, but most of all it gives information on were the fuel is located. The fuel density will have a direct impact on the possibility for using spark assistance as well as for the auto ignition behaviour and location.

21

Figure 11. LIF sequence of SACI combustion. This work is conducted in collaboration with the combustion physics department, performed by Håkan Persson, Johan Sjöholm and Elias Kristensson with the support from Bengt Johansson, Mattias Richter and Marcus Aldén.

Published results H. Persson: ”Spark Assisted Compression Ignition (SACI)-Trapped Residuals and Optical Experiments”, Licentiate Thesis, Lund 2006, ISRN LUTMDN/TMHP— 06/7041—SE

Results to be published H. Persson, A. Rémon, A Hultqvist, B. Johansson: ”Investigation of the Early Flame Development in Spark Assisted HCCI Combustion Using high Speed Chemiluminescence Imaging”, Accepted for publication in the SAE 2007 World Congress H. Persson, A. Rémon, B. Johansson: “The Effect of Swirl on Spark Assisted Compression Ignition (SACI)”, Submitted to 2007 JSAE/SAE International Fuels and Lubricants meeting

22

HCCI Combustion Control Persons involved The combustion control project has involved 2 senior researchers; Associate Professor Per Tunestål and Professor Rolf Johansson, 2 PhD students; Carl Wilhelmsson and Anders Widd and one master student Raphael Vargas. Rolf Johansson and Anders Widd from the Department of Automatic Control have joined the project during the year which means that most of their efforts have been spent on getting up to date with the work within the project. Per Tunestål and Carl Wilhelmsson from the combustion engine division have continued their activities within the project whereas Raphael Vargas has done his master thesis work within the project as a visiting student from Spain. Engine experiments have been performed on the Scania D12 6-cylinder engine.

Publications FPGA Based Engine Feedback Control Algorithms Wilhelmsson, Carl and Tunestål, Per and Johansson, Bengt, 2006, Published in: FISITA 2006 World Automotive Congress Model Based Engine Control Using ASICs: A Virtual Heat Release Sensor Wilhelmsson, Carl and Tunestål, Per and Johansson, Bengt, 2006, Published in: Les Rencontres Scientifiques de l'IFP: "New Trends in Engine Control, Simulation and Modelling" The Performance of a Dual Fuel HCCI Engine Raphael Vargas, Master Thesis 2006 at Lund University, Faculty of Engineering

Experimental setup Engine mapping experiments have been performed on the Scania D12 6-cylinder engine with natural gas and n-heptane as fuels. A VGT turbo was added to the engine system to provide another degree of freedom to the charging system (see Figure 12).

23

INTER COOLER TEMP VALVE Q

AIR FLOW METER

HEATERS

n-C7H16 NG

VGT TURBO

ENGINE

Figure 12: The Scania D12 6-cylinder engine converted to HCCI operation. FPGA experiments have been performed on the same desktop system as the previous year (see Figure 13).

Figure 13: Experimental setup for the virtual heat release setup using FPGA technology.

24

Studies performed Mapping and operating strategy with NG / n-heptane and VGT turbo This engine mapping was mainly carried out by MSC student Raphael Vargas during early 2006 and was reported in a corresponding MSC report. The engine which is connected to KCFP control activities (Dual fuel Scania HCCI) were evaluated in a as large part as possible of its operational regime. Before the start of this work the engine were fitted with a variable geometry turbo as an attempt to increase its operating range. The major limiting factor for this experiment was the emissions of NOx, which was limited to 0.2g/kWh according to US2007 emission regulation. In order to set the operational parameters in a well defined manner an operational strategy were developed after some initial tests. The aim of the operational strategy was to link the different operational parameters with the help of some operational rules. Two datasets were obtained with the engine operated according to these rules, one with as low boost pressure as possible and one with the boost pressure set (according to one of the rules) to maximize net indicated efficiency, still maintain NOx emissions within limits (0.2g/kWh). Data from the two different datasets were thoroughly analyzed and presented in the MSC report. The MSC report did focus on the performance and emissions in different parts of the operational map. A paper was also written (JSAE 20077035) and will be presented at the Fuel and Lubes conferences in Kyoto Japan during 2007. The main topic of the paper are moved away from the pure performance in different parts of the operational regime towards the actual strategy used to select operational parameters the development of the same and the outcome from the use of the strategy. This topic is regarded to offer more “news value”. Among the conclusions that are to find in the paper it can be found that:

Figure 14: NOx emissions in the complete operational area (non boosted operation on the left). The deployment of the turbo increased the maximum load from 4 to 6 bar IMEP within the selected maximum emission level, see Figure 14Figure 14: NOx emissions in the complete operational area (non boosted operation on the left). A net indicated efficiency of 45-50% can be obtained. Some efficiency decrease compared to a naturally aspirated engine would have to be accepted in order to suppress NOx formation in the upper load points. Natural gas seems to be an ill-suited fuel for HCCI engines. The high-load limit of the operating range was due to excessive NOx

25 emissions and one reason is the small temperature span available with natural gas between auto ignition and NOx formation.

Virtual Heat Release Sensor with FPGA Technology The efforts of the control PhD student did during 2006 mainly belong to the effort in designing the virtual, FPGA based, Heat release sensor. Due to large issues of practical nature the work of developing the final version of the heat release sensor progressed more slowly than expected after the initial tests of 2005. Finally however the goal of a working virtual heat release sensor succeeded. A prototype version was developed and is operational on the desktop experimental setup visible in Figure 13. A successful design was finished in the late spring of 2006 and the results were presented at the 2006 FISITA world automotive congress held in Yokohama, Japan, late October. The results were also represented as a poster on the, Published in: Les Rencontres Scientifiques de l'IFP early October. The virtual sensor itself consists of a Field Programmable Gate Array mounted on an evaluation circuit board. This evaluation circuit board were fitted with AD and DA converters to enable the FPGA to operate on analogue signals. The FPGA, AD and DA converters featured state of the art performance, never the less the total system cost were to regard as negligible (roughly 700 euro). For the design of the logic residing within the FPGA, a Xilinx program (System Generator DSP) in combination with Matlab/Simulink was used. The development system can best be described as “rapid prototype system” and enables the designer to set up, design and test the design in a simple way. The work in designing the logics was time demanding. Treating and calculating data based on cylinder pressure signals at the throughput frequencies in question are a very difficult task. The input circuitry of the AD converters had to be thoroughly modified to be able to handle the (in this perspective) very low frequency cylinder pressure signals. The “standard” heat release algorithm also had to be re-written to better suit the algorithmically environment within the FPGA, this reformulation are apparent from Equation 1, please note that Equation 1 is completely equivalent to the “standard” heat release equation, it is only formulated differently.

Equation 1: the reformulation of the “normal” hetrelease equation.

26 When the initial issues finaly were overcome the performance of the virtual heatrelease sensor proved to be impressive. The throughput of the system are 50MHz, and the “calculation time” meaning the time it takes for one heat release sample to be calculated from the instance when the corresponding cylinder pressure sample are available to the circuit is 120ns. Compared to the time base of the actual combustion engine (which at 1200rpm are able to revolve 0.000864CAD in the same time) the heat release calculation time is neglect able, hence the name “virtual heat release sensor”. This moves the time horizon for feedback combustion control from “cycle to cycle” control to “in cycle” control. An indication of true in cycle performance is given in Figure 15 showing two consecutive cycles, the first cycle is a simulated motored cycle; the second is a simulated fired cycle. Several consecutive cycles, both motored and fired are shown in Figure 16. Measured emulated Pcyl

6

8

x 10

Pcyl [Pa]

6 4 2 0

400

500

600

700 800 900 CAD [°] Corresponding Q measured from FPGA

400

500

600

1000

1100

1000

1100

Q [J]

1500 1000 500 0 −500

700 800 CAD [°]

900

Figure 15, two consecutive simulated cycles, note the response of Q in the same cycle.

27

Measured emulated Pcyl

6

8

x 10

4

P

cyl

[Pa]

6

2 0 320

330

340

350

360 370 380 CAD [°] Corresponding Q measured from FPGA

390

400

330

340

350

390

400

2000

Q [J]

1500 1000 500 0 −500 320

360 CAD [°]

370

380

Figure 16, multiple cycles, the spread between the cycles is caused by issures on the input circuitry of the AD converter. Among the conclusions of the paper taking account for these experiments can be read; disregarding the synchronous thinking between the engine and FPGA board enables very low latencies without any major drawback. Extraordinary performance can be achieved using the described system: A sample of cylinder pressure Pcyl is processed and corresponding QnetHR is calculated before any engine can move 0.02 CAD. A throughput of 50 MHz enables the FPGA system to perform 12 120*5 sample HR analyses within a single CAD @ 1200 rpm. Further performance improvement can easily be obtained through the use of more powerful hardware.

Self-Tuning Cylinder Pressure Based Heat Release Accurate tuning of heat release models is difficult and time consuming. Furthermore it requires re-tuning for different operating points which more or less makes it unfeasible for implementation in an ECM featuring cylinder pressure based closedloop combustion control or advanced cylinder pressure based OBD. A new self tuning heat release algorithm has been developed within the Center of Competence for Combustion Processes that estimates the polytropic exponent and pressure sensor offset during the compression and expansion strokes. These parameters are interpolated during the combustion event and then an apparent heat release calculation is performed using the interpolated parameters. The result is a well tuned heat release calculation that re-tunes itself for every new pressure cycle.

28

The estimation algorithm is a Newton based least-squares algorithm that takes advantage of the special structure of the estimation problem with linear dependence on the pressure sensor offset and nonlinear dependence on the polytropic exponent. The result is very fast convergence with a relative accuracy of 1 ⋅ 10 −4 for the polytropic exponent within only four or five iterations. Figure 17 shows the difference between heat release computed with the suggested algorithm and apparent heat release based on a constant polytropic exponent. The rapid convergence makes the algorithm highly suited for ECM implementation as well as online test cell indication.

Figure 17: Heat release based on new algorithm (solid) and constant polytropic exponent (dashed) respectively.

More information about the algorithm and its characteristics can be found in the report “Self Tuning Cylinder Pressure Bsed Heat Release Computation” by Per Tunestål.

29

Modeling of HCCI Combustion Personnel This work is conducted by Ph.D. students Rixin Yu and Tobias Joelsson under supervision of Professor Xue-Song Bai.

Background This is a new project started from January 2006. The work is motivated by the need of deeper understanding of HCCI and PPC combustion processes in engines. Experimental studies on these processes have achieved great success in the past phases of the KC-FP program. Modeling of the chemical kinetic aspects of HCCI combustion has also been carried out within the KC-FP program. Previous modeling activity had been limited to low spatial dimensional stochastic reactors. Experimental results and chemical kinetic modeling suggest that HCCI combustion is greatly affected by the in-cylinder turbulence field. A stronger turbulence field, generated by for example piston bowl geometry, could cause earlier ignition of the charge, longer combustion duration, and slower pressure rise-rate. This could be useful for developing control strategies in the future HCCI engine development. However, there is a lack of understanding of the fundamentals of turbulence/chemistry interaction in the HCCI combustion process. This is to certain extent due to the lack of modeling and diagnostics of the three dimensional auto-ignition kernels, and the stratified temperature field. In the present modeling project we have adopted the spatial and temporal resolved large eddy simulation (LES) method to simulate the onset of the temperature inhomogeneity and the development of flame kernels.

Investigation of the temperature stratification and in-cylinder turbulence LES is carried out to study the basic flow structures in a laboratory diesel (GENDIS) engine, based on a modified light-duty VOLVO D5 engine. PIV measurements were carried out for the same engine previously in the GENDIS project. The aim of the LES modeling was to gain deeper understanding of the formation mechanisms of incylinder turbulence and temperature stratification in different stages of the piston motion, by capturing the fine turbulence structures in the cylinder, and the turbulent heat transfer field. From the LES and the PIV data it is evident that turbulence production in the intake and compression phases is mainly due to the large scale flow motion. As shown in Figure 18 and Figure 19, in the intake phase, the tumble and swirl motions introduce strong large-scale vortex pipes in the shear layers. These large vortex structures break down and thereby transfer energy from the large-scale (nonturbulent) structures to the smaller turbulent eddies. The large-scale vortex pipes are found near the intake valve and also near the corner of the piston surface and bore walls, as well as near the rim of the piston bowl. In the later stage of the compression phase the flow is characterized by more isotropic turbulent eddies. Turbulence vortex cores exist mainly in the engine bowl. In the expansion phase, turbulence vortex cores are also mostly generated near the shoulder of the piston bowl, particularly during the earlier stage of the expansion phase when the piston is near the TDC. This demonstrates the important impact of the geometry of these engine parts on the turbulence field. Since turbulence controls the mixing of fuel and air in Diesel engines

30 and the flame propagation process in SI engines, it is important to understand the development of turbulence in the engine cylinders. A

B C

D

A

B D C

Figure 18. Instantaneous trace lines from LES, showing the flow structures at the earlier intake phase. Top: CAD= −322 ; bottom: CAD= −280 .

Figure 19. Single cycle instantaneous vortex structures from LES, shown by the iso-surfaces of λ2-eigenvalues at different piston positions. The iso-surfaces represent qualitatively the turbulence eddy topology.

31

Figure 20 shows the temperature field at different crank angles. From the LES data it appears that the temperature stratification is mainly due to the heat transfer between the cylinder and piston walls and the bulk flow. During the intake stroke the fresh and cold mixture impinges to the piston and is deflected further towards low flow speed regions. The hot walls in the piston bowl heats up the low speed mixture in the bowl, and the hotter mixture is then further transported by the turbulence eddies to other regions of the cylinder, and this brings about stratification of the temperature field. Meanwhile, the hot bore walls heats up the nearby cold mixture and increases the temperature inhomogeneity. The maximal temperature difference in the cylinder, due to the heat transfer between the cold flow and the hot walls, can be as high as 50 – 60 K. When hot residual gas remaining in the cylinder is considered, the temperature inhomogeneity is even higher. The present findings are useful for understanding the ignition and combustion process in HCCI engines where temperature stratification of the order of 50 – 60 K can significantly affect the ignition timing and combustion duration. More details about the results are presented in paper [1].

Intake inflow (a)

Intake inflow (b)

Intake inflow (c)

Figure 20. A single cycle instantaneous temperature field (in Klvin) development of temperature inhomogeneity in the intake stage. (a) CAD=-322; (b) CAD=-280; (c) CAD=-260.

Effect of Turbulence on HCCI Combustion HCCI combustion process in a 0.5L single-cylinder optical HCCI engine is studied using LES. The engine bore is 81 mm and stroke 93 mm. The engine was equipped with port-fuel injection system, which generates a principally homogeneous charge, and a blend of 50% iso-octane and 50% n-heptane was used as fuel. The compression ratio was set to 12:1 and the engine was run with a lambda (air/fuel ratio) of 3. The engine runs at 1200 rpm. Figure 4 shows snap shots of the formaldehyde (CH2O) and OH-LIF in a cross section of the cylinder at different crank angles. The mass fractions of formaldehyde and OH, the temperature and the flow trace-lines in the corresponding cross section and crank angles, calculated from LES, are also shown in the figure. The development of the reaction/ignition fronts is demonstrated in these images. At CAD = −17 , cool flame chemistry emanates, as indicated by the CH2O field; CH2O develops rapidly as the crank angle increases. In the range −7 < CAD < 2 , CH2O is found in almost the entire cross section shown in the figure. Thereafter, CH2O is consumed and vanishes gradually, and at the same time OH starts

32 to develop and temperature rises in places where OH is developed, indicating that the main ignition has emanated. The LES results and the measurements are in qualitative agreement; in particular both LIF and LES show close correlation of the disappearance of CH2O with the emerging of OH. The interaction of turbulence and the ignition/reaction front is evidenced in Figure 21. The rotational main flow stream is counter-clockwise. The ignition/reaction front is ‘convected’ by the flow stream in the counter-clockwise direction. To elucidate the mechanisms of the interaction of turbulence and the ignition/reaction front propagation and the influence of initial temperature inhomogeneity on HCCI, LES ' ' with different initial turbulence intensity ( u0 ) and temperature inhomogeneity ( T0 ) are carried out. The results are shown in Figure 22. For the same turbulence field, increasing temperature stratification leads to an increase of the maximum in-cylinder pressure. The ignition process is enhanced by the increase of initial temperature inhomogeneity. The influence of turbulence on the ignition is not monotonic. For a moderate and low temperature stratification, increasing the initial turbulence intensity leads to a decrease of maximum in-cylinder pressure. The ignition process is suppressed by the increase of initial turbulence intensity. However, for a high initial temperature inhomogeneity, increasing the initial turbulence intensity leads to an increase of the maximum pressure. The influence of turbulence in this case is to prompt the ignition. More details about the results are presented in paper [2]. Exp. grey=CH2O, red=OH CAD

-17

-7

+2

+10

+15

Figure 21. First column from left: Single-shot PLIF images from onset of cool flame first stage ignition until the end of the main ignition. Formaldehyde is shown in grey and OH is shown in red (dark grey in grey scale). Last three columns from left: LES predicted instantaneous distribution of formaldehyde (CH2O), OH, temperature and fluid trace lines for initial condition u0' = 2.85 m/s,

T0' = 70 K.

Cylinder Pressure [bar]

33

50

u'=0.57 0 u'=1.20 0 u'=2.85 0

45

u'=4.00 u'=5.70 Exp.

40

35

30 60

80

100

T'0 [K]

(a) u'0 =0.57 u'0 =2.85 u'0 =5.70 u'0 =0.57 u'0 =2.85 u'0 =5.70

20

CA10 CA10 CA10 CA50 CA50 CA50

CA10, CA50

15

10

5

0

-5

40

60

80

100

T'0 [K]

(b) Figure 22. (a) Peak in-cylinder pressure calculated from the LES under different initial turbulence intensity and temperature inhomogeneity conditions. (b) Crank angle with 10% heat release (CA10) and crank angle with 50% heat release (CA50) calculated from LES under different initial turbulence intensity and temperature inhomogeneity conditions.

34

Future Work In 2007 LES is to be carried out to study the HCCI combustion process of ethanol/air mixture in an experimental optical engine. Experimental studies on this engine were carried out in 2006. The engine is a six cylinder Scania D12 truck sized diesel engine that was converted to HCCI operation by the use of port fuel injection (PFI). The combustion was made in a single cylinder whereas the rest cylinders were motored. Two different piston shapes, one with a flat disc and one with a square bowl, were employed to generate different in-cylinder turbulence and temperature field prior to auto-ignition. It was shown from the experiments that the piston bowl has an earlier ignition in the bowl, higher combustion duration and slower pressure rise rate as compared to the disc engine. This is likely due to temperature stratification as we have shown in the previous results. However, measurement of the temperature stratification is difficult and unavailable for the present engine. The aim of this study was to scrutinize the effect piston geometry on in-cylinder turbulence and the temperature field as well as the combustion process.

Publications 1. Yu, R., Bai, X.S., Hildingsson, L., Hultqvist, A., Miles, P.: Numerical and Experimental Investigation of Turbulent Flows in a Diesel Engine, SAE paper 2006-01-3436, 2006. 2. Yu, R., Bai, X. S., Lehtiniemi, H., Ahmed, S. S., Mauss, F., Richter, M., Aldén, M., Hildingsson, L., Johansson, B., Hultqvist, A.: Effect of Turbulence and Initial Temperature Inhomogeneity on Homogeneous Charge Compression Ignition Combustion, S AE paper 2006-01-3318, 2006.

35

SI Gas Engine Project Persons involved The gas engine project has involved one senior researcher, Associate Professor Per Tunestål, and one PhD student, Mehrzad Kaiadi. Both are with the combustion engine division and have been active in the project for the full duration of 2006. Since the project has just been started, most of the year has been spent on literature review and planning of research activities.

Publications No publications have been published during 2006.

Experimental setup A new natural gas engine from Volvo Powertrain has been installed in the engine lab. The engine has recently been motored and the first operation with combustion is planned during the first quarter of 2007. An open control system for fuel injection and spark control from Mecel Engine Systems is just about to be delivered. The control system from Mecel Engine Systems will accept fuel and spark commands over a CAN bus from a controlling computer (see Figure 23). The Mecel system is also capable of measuring ion current.

CAN bus

Mecel modules

intake

Linux computer

Engine

Figure 23: Engine control system layout.

The controlling computer will run the Linux operating system and a C-program that takes care of measurements and communication with the Mecel system. Controllers will be designed using Matlab’s Simulink and Real Time Workshop will be used to convert the control designs to C-code. A similar setup is in use for HCCI control in another project and presently the setup is being migrated and converted for the natural gas engine setup.

36

Studies performed Only literature studies have been performed so far.

Planned activities Ultra-diluted SI operation with dilution limit control based on ion current or cylinder pressure measurements. Transient control strategies will also be developed.

37

KCFP subproject: Fuel Effects Personnel Rolf Egnell, Associated Professor. Ph.D, and Uwe Horn Ph.D. Student.

Introduction After a very long period of stability and continuity on the fuel market it is now obvious that the expected shortage of crude oil, our concern of global warming and recent progress in engine development will change our view on fuels for combustion engines. A reduction of oil production will increase our demand for fuels to replace the conventional fuels. To keep down the anthropogenic expels of green house gases there is a need for fuels derived from CO2- neutral feedstock. The increased knowledge of the interaction between the fuel, the combustion process, efficiency and emission formation in combustion engines give rise for new and more specific demands on the fuels. In addition, new combustion processes for IC-engines have evolved that call for new means for characterisation of fuels. The Fuel Effect project was initiated at the beginning of the year 2006 and the Phd student Uwe Horn was employed in March the same year. The research work performed by Uwe during his first year is summarized below.

Fuel Specific Differences between Euro Diesel Fuel and RME under partially premixed combustion In the degree that costs and demand of crude oil rise, diminish the economical disadvantages for alternative Diesel fuels, resulting in a variety of feasible substitutes. Diesel fuel substitutes have deviating exhaust emissions from conventional fuel. The methyl ester of rapeseed oil (known as RME/Biodiesel) is receiving increasing attention as an alternative fuel for Diesel engines. RME is a non-toxic, biodegradable, and renewable fuel with the potential to reduce engine exhaust emissions [1]. The main disadvantage for RME is its vaporization and self ignition characteristics at low load conditions. Engine experiments were carried out at 4 bar IMEP with Euro Diesel fuel (EDF) and RME. During these engine experiments EGR and injection pressure were varied. As a result, differences in exhaust emissions due to EGR, injection pressure and fuel type were observed. The objective of this work was to find answers for fuel dependent differences in indicated load and exhaust gas emissions. As combustion and emission formation of RME has not been fundamentally explained yet [2], a detailed analysis approach based on explanation models for fuel characteristics was chosen to explain the observed differences.

38 FUEL CHARACTERISTICS

SAUTER MEAN DIAMETER Spray vaporisation properties for RME deviate considerably from EDF. This is due to increased fuel viscosity and surface tension which entail larger average droplet size. A measure for the average droplet size is the Sauter Mean Diameter (SMD): The strongest parameters influencing SMD are the velocity difference between fuel and cylinder charge ( ug ), the fuel viscosity and fuel surface tension. As ug is highly dependent on injection pressure, it is presumed that SMD is considerably influenced by increasing injection pressure. DISTILLATION CURVE Fuel composition is indicated by the shape of the distillation curve [3]: The cetane number of the fuel fraction vaporising during the start of the distillation curve has an influence on exhaust emissions: A low initial fuel cetane number (ICN) increases ignition delay. Hence, both PM and NOx-emissions are lower with ICN improvers even though the overall fuel cetane number is virtually unaffected.

Distillation Fraction [-]

1 0.8

EDF RME

0.6 0.4 0.2 0

200

250 300 350 Distillation Temperature [°C]

Figure 24: Distillation curve for RME and European Diesel Fuel EDF

A flat distillation curve is advantageous due to a lower fuel evaporation rate over the whole distillation temperature interval. It has been reported that thermal engine efficiency increases due to a better controlled vaporized fuel supply during the whole combustion process [3]. Even though the CN for RME is slightly higher than for EDF, combustion properties are reported to be poor at low load conditions [3,4,5]. This is due to the composition of RME: RME consists of a few types of acid ethyl esters (C16 – C20, thereof 90% C18) with similar properties whereas EDF consists of 200 different hydrocarbon types (C9-C16) with varying properties. Hence, RME can be considered as a virtually pure substance with a very high distillation curve gradient (cf. Figure above – Distillation Temperature): Especially under low load conditions, fuels with a strong distillation temperature gradient cause HC-, CO- and PM-emissions. HC- emissions are produced by incomplete combustion due to poorly controlled fuel vaporisation as a result of a short distillation temperature interval.

39 The initially vaporised fuel fraction has a major influence on ignition delay. Thus, a high boiling point increases ignition delay as more time is needed to vaporise the initial fuel fraction. For the differences in fuel characteristics between EDF and RME, the following is summarised: • •

EDF (certification fuel) was chosen as the reference fuel. It contains ignition delay improvers with low ICN and oxidation characteristics are improved by a low volatile fuel fraction with high FCN. For RME, fuel properties as viscosity, heat capacity, boiling point, distillation curve, volatility and chemical structure are different from EDF. This means that formation and growth of particulates during combustion as well as oxidation characteristics differ considerably.

RESULTS

Influence of EGR and Fuel Type

4.5

CO [%]

250 0.75 0.5 0.25

0

0

750 500 250

FSN [-]

2

0

3

750 500

EDF RME

HC [ppm C ]

4

NOx [ppm]

IMEPgross [bar]

Emissions changed during variation of EGR under constant fuelling rate. Fuelling rate was adjusted for differences in density and heating value in order to inject the same amount energy for both fuels. Fuel specific differences in indicated mean effective pressure of the combustion phase (IMEPgross) and exhaust emissions were observed. Exhaust emissions converged for both fuel types with increasing EGR level.

1 0 0

10

20

30 40 EGR [%]

50

60

Figure 25: Effect of EGR.

These observed differences are explained by comparison of the rate of heat release (HRR) between 0 and 50% EGR: As shown in the figure below both, premixed and diffusive HRR is lower for the RME case. At low load conditions, RME is considered as a poorly ignitable diesel fuel [3 pg. 125ff] which is due to a high distillation middle temperature (T50): If it is assumed that the overall liquid fuel spray has to be heated

40 until T50 to become vaporised, and if it is moreover assumed that heat capacities for both fuel types are alike, the heat needed for warming up the liquid spray phase to vaporisation conditions is higher. Due to the high distillation temperature gradient, more time is needed for vaporisation of the initial fuel fraction, which increases ignition delay somewhat. 0...20 %EGR

30...50 %EGR

200

0

increasing EGR

Premixed →

0.4

0.6

0.8

0.4

0.6

0.8

← Diffusive

200 100

2

D2HR [J/°CA ]

0 50

0.4

0.6

0.8

EDF 0.4 RME

0.6

HRR [J /°CA]

HR [J]

400

0.8

0 -50 0.4 0.6 0.8 0.4 0.6 0.8 time after SOC [ms] time after SOC [ms]

Figure 26: Combustion process

Due to RME’s mono component like short distillation temperature interval a large fuel spray fraction vaporises concurrently within a small temperature and spatial interval if enough ambient heat is available. Together with a larger average droplet size due to higher fuel viscosity increase these fuel characteristics the affinity of RME to form a higher amount of fuel rich reaction zones at start of combustion. This is reflected in a lower overall HR due to poor premixed and diffusive combustion (see figure above). Both, overall and premixed HR is delayed for the RME case between 0...20% EGR which is in accordance with a decrease in maximum HRR. If EGR level is increased above 30%, the point of maximum HR is more delayed for the EDF case which is in contrast to the 0…20% EGR cases. This behaviour is related to the ignition delay which increases with EGR level. The time delay between SOC and maximum HRR (HRRmax – see figure above) is a function of mixture quality; hence, the ignition delay for the RME case at EGR levels above 30% is an indicator for improved mixture generation. Ignition delay of RME was similar to EDF. For EDF, the ICN is lower than the overall CN to increase ignition delay. For RME against that, the ICN is equal to the

41 overall CN. The ICN is considered to have the strongest influence on ignition delay. Hence, the similar ignition delay between EDF and RME is related to a longer evaporation process for the RME-case; more time is needed to form an ignitable fuelair mixture. For the influence of EGR on different fuel types, the following is concluded: For the RME case, overall HR as well as the maximum HRR is lower than for EDF. For EGR levels above 30%, combustion phasing was slightly earlier than for the EDF-case which was in contrary to low EGR levels. This is an indication for an improvement in mixture generation for the RME case with higher EGR levels due to increased ignition delay. Influence of Rail Pressure The variation of rail pressure led to a further indication for the assumption of locally fuel rich combustion of RME as the source for PM. Observed differences in HR for EDF and RME were considerable:

4.5

EDF RME

350 200

3

500

HC [ppm C ]

4

0.3 25 20 15 10

FSN [-]

2

NOx [ppm]

0.2

1 0 75 25 500

600

700 800 pRail [bar]

900

SMD [μm]

CO [%]

IMEPgross [bar]

A decrease in ignition delay and combustion duration was observed, while HRmax and HRRmax increased concurrently with higher injection pressure. Increased HRmax is an indication for improved mixture generation. This indication is emphasised by calculation of average droplet size. In the following picture is depicted, that SMD and exhaust emissions are considerably influenced by injection pressure:

1000

Figure 27: Effect of injection pressure

Injection velocity is the strongest factor influencing SMD. Increasing injection pressure increases relative velocity between injected fuel und cylinder charge. Hence, average droplet size is reduced which improves vaporisation characteristics for all fuels. RME has a larger SMD due to higher fuel viscosity and benefits stronger from increasing injection pressure. This benefit is reflected in decreasing emissions and

42 increasing IMEP. For the influence of increasing injection pressure on HR, the following is summarised: • Average droplet size decreases with injection pressure, which improves fuel spray evaporation characteristics. •

•

Improved fuel mixture generation is reflected in improved HRmax, shorter ignition delay and shorter combustion duration. The fuel type RME benefits stronger from increased injection pressure than EDF. This is related to a stronger reduction in SMD compared with EDF.

CONCLUSIONS

Measurement results from a HSDI Diesel engine were studied by detailed heat release analysis. For both fuel types, increasing EGR levels increase ignition delay, which improves mixture generation and premixed combustion. Increasing EGR level entails a longer premixed combustion duration, which lowers maximum HRR. In this way the amount of fuel rich, locally hot zones is minimised. Hence, PM emissions originating from fuel rich combustion were improved. RME showed different combustion characteristics from EDF at low load conditions which depend on the following factors: •

• • •

•

Injection duration was longer for the RME case due to lower fuel energy content. This resulted into less available time for fuel mixture generation; a larger fuel fraction was injected close to or after SOC, which influenced HRR at low EGR conditions. For RME, the high ICN resulted into early ignition of fuel rich zones entailing suboptimal premixed HR which was indicated by lower maximum HR. Due to higher fuel viscosity, SMD is increased for RME. Increased average droplet size increases the affinity to form locally fuel rich zones. Fuel dependent SMD differences are compensated by increased injection pressure. RME combustion characteristics can be considered as poor at low load conditions. Due to a high distillation middle temperature, short distillation temperature interval and increased SMD, fuel vaporisation is delayed. Premixed HR is unsteady due to concurrent fuel spray evaporation and premixed combustion. The ratio of diffusive and premixed HR is lower for RME than for the reference fuel. This is related to the lack fuel components that improve FCN. The lower diffusive HR entails poor emission oxidation characteristics.

If alternative diesel fuels with varying properties shall be included in future emission legislations, technical possibilities exist to meet these legislations: Today’s Diesel engines are equipped with advanced combustion control devices like EGR, Common Rail and Turbo Charging. In addition, fuel and/or cylinder pressure sensors open possibilities for a multi fuel production engine. If engine control strategy is adapted to a specific fuel type during operation, emission demands can be met for a variety of fuels.

43

Future work •

•

•

As some reported phenomena as the load dependency of both IMEP and emissions during combustion of methyl esters has not been completely explained yet, the next step within the fuel project is to do engine measurements at different load conditions, external EGR and injection pressure. The influence of different ME feed stocks shall also be evaluated by using both, RME and SME In cooperation with the GenDies group it is planned to do measurements with a mixture of N-heptane and iso-octane. The purpose of this study is to obtain further knowledge on fuel parameters by using “laboratory fuels” and to do comparative measurements with the optical engine from the GenDies lab to explore HR differences due to the optical measurement setup. Afterwards the possibility of methanol as a fuel during PPC combustion shall be evaluated. Self ignition of methanol shall be obtained with the following measures: o Increased compression ratio o Excessive intake air heating o Glow plug assistance o Warm external EGR o Internal EGR by NVO

•

o Lubricants (to avoid problems with the CR-injection system)

Depending on the results obtained from the methanol engine tests it will be decided how to proceed with the fuel project.

References 1. Hopp, M.: “Untersuchung des Einspritzverhaltens und des thermischen Motorprozesses bei Verwendung von Rapsöl und Rapsmethylester in einem Common-Rail-Dieselmotor“, Dissertation Universität Rostock 2005 2. R. L. McCormick, C. J. Tennant, R. R. Hayes, S. Black, J. Ireland, T. McDaniel, A. Williams, M. Frailey: “Regulated Emissions from Biodiesel Tested in HeavyDuty Engines Meeting 2004 Emission Standards”, SAE 2005-01-2200 3. Garbe, T.: “Senkung der Emissionen eines PKW mit direkteinspritzenden Dieselmotor durch Verwendung von Kraftstoffen mit abgestimmtem Siede- und Zündverhalten”, Dissertation Universität Hannover 2002 4. Chang, D., Van Gerpen, J.: “Determination of Particulate and Unburned Hydrocarbon Emissions from Diesel Engines Fueled with Biodiesel”, SAE 982527 5. Mayer, A.C.R., Czerwinski, J., Wyser, M.: “Impact of RME/Diesel Blends on Particle Formation, Particle Filtration and PAH Emissions” SAE 2005-01-1728

44 PUBLICATION

Horn, U., Egnell, R., Johansson B., Lund Institute of Technology and Andersson, Ö., Volvo Car Corporation “Detailed Heat Release Analysis With Regard To Combustion of RME and Oxygenated Fuels in a HSDI Diesel Engine”. Accepted for publication at the 2007 SAE World Congress.

45

The Generic Diesel Research (GenDies) project 1. Introduction Diesel engine developers are facing major challenges when meeting future demands on NOx and particulate emissions. The exhaust after treatment systems are rapidly improving and the technologies available in a few years will be capable of cutting emission levels close to the demanded levels. However, in order to reach the low emissions required in a long-lasting, cost effective way, the potential of reducing engine-out emissions must be fully utilized. Fortunately, engine management systems and hardware are now available to optimize fuel injection and heat release for low emissions while maintaining the trademarks of the diesel engine, i.e. low fuel consumption and dependable performance. The possibilities offered by the new technologies are so wide-ranging that, in order to use them optimally, models must be developed to guide the engine developer. The most advanced models, the Computational Fluid Dynamic (CFD) models, are now being established as useful tools in the Diesel engine development. However, the development and success of these models are critically dependent on the possibility to validate them in a realistic environment. The main purpose of the Generic Diesel engine research project (GenDies) is to create an experimental environment for validation of advanced CFD-models and for studying basic phenomena and futuristic concepts of common interest. Thus, the work is focused on increasing the general knowledge of the combustion in diesel engines, generation of data sets and development of advanced models, i.e. pre-competitive research rather then solutions for production engines. The GenDies project has been going on since 2002, initially as a part of the Center for Combustion Sience and Technology (CeCost) in Lund, and since 2006 as a part of the KCFP. This report covers the work performed during 2006. For the sake of continuity some results from some activities from 2005 are also included. The GenDies subprojects reported below are: •

•

•

Diesel engine flow measurements, Combined measurements of flow field, partially oxidized fuel and soot and, and temperature and heat flux measurements Optical Analysis of Soot Reduction by Post Injection under Dilute Low Temperature Diesel Combustion Study of Flame Lift-Off Length in Relation with Engine-out Soot in Heavy Duty Diesel Engine.

46 •

•

Comparison between optical and all metal engines High-speed video campaign for measurements of simultaneous OH and natural luminescence and temperatures measurements with two-color pyrometry

2. Gendies staff Rolf Egnell Mattias Richter Robert Collin Leif Hildingsson Ulf Aronsson Clemént Chartier Combustion physics PhD students

Jan-Erik Nilsson

Project leader, combustion engines (since Sept 2006 succeeding Anders Hultqvist). Senior researcher, combustion physics Senior researcher, combustion physics Senior researcher, combustion engines PhD student, combustion engines PhD student, combustion engines PhD student, combustion physics, varies depending on employed measurement technique Technician, combustion engines

3. Diesel engine flow measurements, Combined measurements of flow field, partially oxidized fuel and soot and, and Temperature and heat flux measurements The following staff was active during the measurements and analysis in this subproject Anders Hultqvist Leif Hildingsson Paul Miles Mattias Richter Robert Collin Jan-Erik Nilsson

Project leader, combustion engines PhD student, later senior researcher, combustion engines Researcher, Sandia National Laboratories Senior researcher, combustion physics Senior researcher, combustion physics Technician, combustion engines

3.1 Diesel engine flow measurements The world’s first PIV measurements in a fired diesel engine were performed within the Gendies project. The investigations were carried out using the personal car diesel engine in the Gendies project which was fitted with an optical piston with a bowl-inpiston-design. Two low-temperature operating concepts were investigated, one similar to a Toyota/AVL concept using quite a lot of EGR, inlet [O2] = 12%, and with conventional timing, SOI = -12.5°. The other operating condition was similar to the Nissan MK concept with late timing, SOI = -4.0° and inlet [O2] = 15%. Figure 28 shows the associated heat release rates.

47

Figure 28: Rate of heat release for the late injection timing case (SOI = -4°) to the left and for the conventional timing case (SOI = -12.5°) to the right.

In-cylinder mixing is expected to be important because of the high EGR levels used as well as the reduced time available to mix compared to strategies with very early injection timing. Despite the importance of the in-cylinder flow structures, very little experimental information exists on them from fired diesel engines. Those that existed until the present investigations were limited to a few points or a traversing a measuring point to provide a profile. However, no planar measurements, as those provided by PIV, existed. As the fuel jet hits the bowl wall and is redirected it sets up a standing vortex in the bowl for both operating conditions. For the conventional timing case by 10 CAD, a reverse-squish flow has developed, Figure 29. Close to the cylinder centreline, a small counter-clockwise rotating vortex has formed which moves outward between 15 and 20 CAD and dominates the upper part of the bowl. As it moves outward it imparts an upward looping motion to the reverse-squish flow. Because of this looping motion near the bowl rim, fluid entering the squish region comes from the uppermost regions of the cylinder. Traces of the looping reverse-squish motion can be seen as late as 60 CAD, but after 45 CAD the structure dissolves to a great extent. For the late injection operating condition, some flow directed into the squish volume is apparent as early as 10 CAD, though it is not as pronounced as is seen with conventional injection timing. As heat release becomes significant around 15 CAD, Figure 30, this reverse-squish motion strengthens, and is further emphasized at 20 CAD. At this time, the reverse-squish motion is considerably stronger for this operating condition than for the conventional injection case, which is consistent with the later combustion phasing. What can also be noted is that transport of combustion products into the squish volume occurs mainly from the outer parts of the bowl for the conventional injection case but for the late injection condition the flow leaves the bowl mainly from the mid-part of the bowl.

48

Figure 29. Mean flow field at different crank angle positions for the operating condition with conventional timing (SOI = -12.5°), 12% O2 concentration. Half bore is imaged.

Like the initial, strong clockwise-rotating vortex formed just after the fuel injection event, the formation of the smaller, counter-rotating vortex near the near the cylindercentre is delayed for the late injection operating condition. However, by 15 CAD this vortex has formed. The counter-clockwise motion grows outward and drives a portion of the flow up into the reverse squish volume but this vortex does not dominate the bowl flow as thoroughly as was seen with conventional injection timing. By 25 CAD, the central, counter-clockwise rotating vortex has combined with the reverse squishflow to form a distinct mushroom-shaped flow structure. This flow structure transports the fluid from the bowl into the squish volume but also towards the cylinder centre. A similar shape can also be observed in the conventional injection case; however, in that case the flow towards the cylinder centre is not as pronounced. By 30 to 35 CAD the mushroom-shaped structure evolves into an upwards looping motion near the bowl rim similar to the motion measured for the conventional injection condition. In contrast to that case, where the looping structure seemed to be attached at the rim, this structure is more like a full vortex. Recognizing the approximate axisymmetry of the in-cylinder flow, this vortex has a toroidal shape

49

Figure 30. Mean flow field at different crank angle positions for the operating condition with late injection (SOI = -4°), 15% O2 concentration, higher swirl

similar to a smoke ring. Although such a vortex is not readily apparent in the conventional injection case, it can be verified to exist by examining the flow in a reference frame moving with the piston; the vortex for the late injection case is more energetic. During this same period, from 30–45 CAD, the flow in the central region of the cylinder is predominantly in the direction of the downward moving piston. No bulk flow structures that might enhance mixing in this region are observed. Due to lower quality data in the bowl for the low swirl case, the examination of the influence of lowering the swirl ratio is restricted to the squish region though vectors in the bowl are presented. Figure 31 shows the flow fields for the late injection case with two different swirl ratios (2.6 and 2.0). The main difference when comparing the images in Figure 31 is the fluid motion entering the squish region. At the higher swirl ratio, this motion has a dominant vertical component and evolves into a toroidal vortex by 30 CAD. In contrast, at the lower swirl ratio, the fluid trajectory is much flatter and the formation of the toroidal vortex is delayed. Additionally, the flow in the inner regions of the cylinder has a stronger component directed radially-inward. These different behaviours are probably due to changes in the location of the heat release. The flatter squish flow trajectory observed near 25 CAD suggests that at lower swirl the main heat release occurs at locations higher in the bowl and closer to the cylinder centreline. Later, near 40 CAD, the inward flow is suggestive of heat release occurring just above the bowl rim. A comparison of the flow structures at lower and higher swirl with conventional injection timing is very similar.

50

Higher swirl

Lower swirl

Figure 31. Flow fields for the late injection case. Higher swirl ratio to the left and lower to the right.

Conclusions from Diesel engine flow measurements • •

• •

• •

Fuel injection event sets up a clockwise-rotating vortex in the bowl region. The trajectory and timing of the reverse squish flow is found to be an important factor influencing the bulk flow motions which likely influence the subsequent mixing of soot and partially-oxidized fuel with unused oxygen Peak reverse squish flow is found to coincide closely with the timing of the peak heat release For the higher swirl cases, the location of the peak maximum in the reverse squish flow are different for the two operating conditions − Conventional timing has maximum close to bowl rim − Later timing has maximum closer to bowl mid-radius Vortex appears near bowl rim just after start of mixing-controlled combustion Looping structure at bowl rim is flatter at lower swirl and the formation of toroidal vortex is delayed. This is probably due to changes in the location of heat release

3.2 Combined measurements of flow field, partially oxidized fuel and soot The operating conditions described in the previous section about flow measurements were also investigated using LII for soot visualization and LIF, with excitation at 355 nm, for visualization of partially oxidized fuel (POF). The operating condition that has been most thoroughly described and most published is the conventional timing case described in the previous section and this operating condition is the one that will be described in this section; intake [O2] was 12%, global air/fuel ratio λ=1.15, load 4 bar IMEP and SOI at -12.5°. Just after injection, the LIF images show little evidence of partially-oxidized fuel (POF) – as is indicated by the mean image obtained at -5°, Figure 32. By -3° POF is clearly observed in the central region of the bowl, where it marks the location of the low-temperature heat release. As the combustion proceeds toward the onset of the main heat release near TDC, POF is still observed in the central bowl, but the highest intensity signal has moved upward to the bowl rim height. The mean velocity field at TDC, lower portion of Figure 32, is dominated by a single vortical structure near the vortex periphery. It is likely that much of the high-intensity POF fluorescence seen at 1° originates from POF which has been transported upward from the central region of the bowl.

51 The divergence, ∇ ⋅ U , can be physically interpreted as the normalized rate of expansion of a fluid element of volume V

− (1 V )( dV

dt ) and it can be estimated

from the two-dimensional velocity fields; in Figure 32 it is represented as a falsecolour background to the velocity fields. Large positive divergence is thus expected to correlate well with high heat release rate. Both simulations and H2CO-LIF measurements in homogeneous charge engines show that a high heat release rate corresponds closely to a high rate of H2CO destruction. Hence, regions of high heat release rate are expected to be found immediately adjacent to regions exhibiting H2CO fluorescence. A similar spatial relationship between the heat release associated with mixing-limited combustion and the locations of PAHs and soot is also anticipated. With this perspective, the divergence observed near TDC is in good qualitative agreement with the distribution of POF. The bulk of the heat release occurs near the edges of the POF distribution, and a major portion occurs high in the bowl, along the anticipated trajectory of the intense POF seen 1° earlier. At 5°, the mean flow divergence indicates that the bulk of the heat release occurs in the outer, upper region of the bowl. Below this region we observe the strongest POF fluorescence. This fluorescence has nearly vanished just 2° later. The disappearance of POF is likely caused by the onset of hightemperature, premixed combustion. Figure 32. POF, velocity field and divergence The first indications of laser- during the initial stages of combustion. induced soot incandescence are observed at 10–11° near the cylinder centreline and, to a lesser extent, near the bowl rim, Figure 33. Temperatures greater than approximately 1700 K are required to form soot on engine time-scales; consequently, the delayed appearance of soot until after significant heat release has occurred is consistent with expected behaviour. The initial distributions of soot are closely mirrored by the distributions of POF – two dominant regions, one near the centreline and the other near the bowl lip. As the cycle progresses, however, little soot is observed outside the centreline region. Because soot is formed in regions characterized by a mixture fraction f greater than approximately twice the stoichiometric value fstoich, the intense LII signal indicates that f > 2 fstoich in % much of the centreline region.

Like the soot LII, the bulk of the POF is observed from the centreline region beyond 11°. In addition to a contribution from soot incandescence, the POF signal at this time likely has a significant component due to PAH fluorescence. However, in contrast to the soot distributions, significant POF fluorescence is observed from the squish volume. It is thus likely that a second fuel-rich region exists within the squish volume.

52 The lack of significant soot formation here may be due to either a less rich mixture or to cooler temperatures than are found in the cylinder centreline region. The dominant heat release zones indicated by the flow divergence estimates are observed in the upper-central bowl region, near the lower portion of the centreline region of intense soot incandescence and POF fluorescence. The vortical structure dominating the flow is positioned such that the fluid in this heat release zone and the POF near the cylinder centreline is either recirculated or extended axially as the piston descends – there is little mean flow motion that could transport this fluid to the outer regions of the cylinder. The bulk of the heat release during the period depicted in Figure 33 appears to be restricted to the bowl. Finally, like the centreline region, beyond 15° there is little mean flow motion in the squish region from which POF fluorescence is observed, other than axial straining. Consequently, the fluorescence is emanating from fluid that was deposited there earlier in the cycle. Much of the behaviour observed at the latter crank angles shown in Figure 33 carries through to the late-cycle period of combustion. Images of soot incandescence show that the soot remains concentrated along the cylinder centreline, as does the bulk of the fluid from which POF fluorescence is observed. Similarly, the spatial distribution of POF fluorescence from the squish volume does not vary significantly. Both regions simply elongate axially, while slowly spreading radially.

Figure 33. Mean flow structures and averaged spatial distributions of soot and partially-oxidized fuel in the latter half of the main heat release and in the initial stages of mixing-controlled burning.

Conclusions from combined measurements of flow field, partially oxidized fuel and soot •

•

The initial distributions of partially-oxidized fuel are found to correlate well with regions of heat release identified from estimates of the mean velocity field divergence. Later in the cycle, soot and partially-oxidized fuel are observed to be concentrated in the central region of the cylinder.

53 •

• •

Within the squish volume little soot is found, although partially-oxidized fuel is clearly observed. Fluid from this region, though not a dominant source of soot emissions, may contribute substantially to unburned hydrocarbon or CO emissions due to the large volume occupied. POF in the outer zone, above the bowl mouth, is located (trapped?) within a toroidal vortex attached to the piston top. The vortex also presents a barrier that prevents fluid exiting the bowl from entering the squish volume. Single-cycle flow fields (and POF/soot distributions) retain many features observed in the mean fields.

3.3 Temperature and heat flux measurements Temperature and heat flux measurements were carried out in the Gendies Scania engine as a part of Tobias Billinger’s master’s thesis work (Diplomarbeit; see publications list). These measurements involved using fast thermocouples to study the influence of EGR and load. Initially, extensive software development work was performed along with characterization and performance measurements of the thermocouples. The measured wall temperature for the EGR-sweep is shown in Figure 34 along with the calculated heat flux. The maximum temperature variation is around 15° which is similar to other investigations. It is remarkable that there is no striking correlation between EGR and temperature level. Due to lower combustion gas temperature with high EGR, wall temperature should accordingly be lower. It is supposed that with already low wall temperatures, effects of different injection timing cover effects of EGR.

Figure 34. Temperature history and heat flux for the EGR-sweep.

Load sweeps with constant intake oxygen concentrations (which leads to different EGR-levels) were also performed, Figure 35. The dip in temperature seen slightly after SOI is not an effect of cooling but of an unfortunate electric disturbance. The figure indicates that temperature history is governed by load, higher load leading to higher wall temperatures and heat flux also correlates well to load.

54

Figure 35. Temperature history and heat flux for load sweep with constant intake oxygen concentration.

3.4 Future prospects, outlook for 2007 and beyond Being able to use advanced heat release analysis along with the optical (including laser) techniques currently in service and those that are being developed there will not be a shortage of possible investigations within the Gendies project. Those studies that are currently under consideration and/or planned for this field of study for the coming period include employing PIV (Particle Image Velocimetry) again, this time to study the effect of injector nozzle protrusion, i.e. the vertical position of the origin of the fuel sprays, on the in-cylinder flow field. It has been reported that even small changes, on the order of one millimetre, in where the fuel spray hits the piston bowl in a diesel engine can have a huge impact on the engine-out emission levels. Thus it would be interesting to find out how the flow field is altered for different injector protrusions. Using the already existing data from the measurements of flow fields and soot and partially oxidized fuel together with new CFD simulations that are to be performed is also underway; this would make for interesting comparison and likely also insights into how the combustion event of partially premixed combustion progresses.

List of publications produced and/or presented during 2006 PhD thesis Leif Hildingsson: “Laser Diagnostics of HCCI and Partially Premixed Combustion” PhD defence on June 2nd 2006

MSc thesis / Diplomarbeit Tobias Billinger: “Development of a Heat Flux Measurement Technique for a Partially Premixed Engine”

Papers 1. Paul Miles, Leif Hildingsson, Anders Hultqvist: “The Influence of Fuel Injection and Heat Release on Bulk Flow Structures in Direct-Injection, Swirl-Supported Diesel Engines”, 13th International Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal 26-29 June 2006

55

2. Henrik Bladh, Per-Erik Bengtsson, Leif Hildingsson, Anders Hultqvist, Volker Gross: “Quantitative soot measurements in an HSDI Diesel Engine”, 13th International Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, 26-29 June 2006 3. Paul Miles, Robert Collin, Leif Hildingsson, Anders Hultqvist, Öivind Andersson: “Combined measurements of flow structure, partially-oxidized fuel, and soot in a high-speed, direct injection diesel engine”, Proceedings of the Combustion Institute, Volume 31, Issue 2, January 2007, pages 2963-2970, Heidelberg, Germany, 6-11 August 2006 4. Leif Hildingsson, Anders Hultqvist, Paul Miles: “The effect of swirl and injection phasing on flow structures and mixing in an HSDI diesel engine”, THIESEL 2006, Valencia, Spain, 12-15 September 2006 5. Rixin Yu, Xue-Song Bai, Leif Hildingsson, Anders Hultqvist, Paul Miles: “Numerical and Experimental Investigation of Turbulent Flows in a Diesel Engine”, SAE 2006-01-3436

4. Optical Analysis of Soot Reduction by Post Injection under Dilute Low Temperature Diesel Combustion Based on the master theses by Clément Cartier

4.1 Introduction Over the past few years, it was demonstrated that dilute low temperature combustion in Diesel engines turns out to be an efficient mean to control NOx and soot emissions, see Figure 36. But these two pollutants are related by a trade-off making it challenging to reduce emissions of both simultaneously. However, experimental efforts led to the observation that post injection of a small fuel quantity can significantly lower the smoke emissions without NOx penalty. But the phenomenons governing this soot reduction are largely unknown.

56

Figure 36: Emissions as a function of EGR

This study investigates the soot reduction mechanism in a diesel engine running under different operating conditions regarding post injection configurations. Finding and understanding of an optimal post injection timing and fuel quantity reducing emissions is the main focus of this work. The effects of post injection on fuel-air mixing are of interest since it is an event occurring in the combustion chamber when combustion has already started. Particle Image Velocimetry analysis is used to investigate the transport velocity of soot particles during the combustion event. The images used during this process are recorded by a high speed camera and feature mainly black body radiation emitted by soot particles. The main injection event is the same for all measurements and the following post injection settings are changed: post injection timing and fuel amount.

4.2 Experimental setup The test engine is derived from the inline six-cylinder heavy duty Scania D12, it is a single cylinder engine modified for optical access.

57 Table 2 gives some of its specifications. The test engine features a Bowditch type optical access to the combustion chamber from below, see Figure 37. All experiments were conducted at 1200 rpm and with an injection pressure of 750 bars.

58

Table 2: Engine Specifications Displacement volume per cyl.

1966 cm3

Bore

127 mm

Stroke

154 mm

Compression ratio Connecting rod

15.5:1 255 mm

Number of valves

4 per cylinder

Inlet valves open

2° BTDC

Inlet valves close

29° ATDC

Exhaust valves open

34° BTDC

Exhaust valves close

6° BTDC

Valve lift Inlet

14.1 mm

Valve lift Exhaust

14.1 mm

Valve diameter Inlet

45 mm

Valve diameter Exhaust

41 mm

Inlet pressure

1.5 bars

Figure 37: A sectioned view of the cylinder (right): (1) glass piston crown, (2) cylinder liner, (3) Bowditch type piston extension, (4) mirror.

59

4.3 Methodology To study the effects of post injection on soot reduction, direct information on performance and pollutant emissions was obtained from engine tests. To obtain a dilute combustion, the EGR rate was kept at 62% for all runs. In order to investigate precisely the effect of post injection, all other parameters were kept constant such as inlet temperature, injection pressure, fuel quantity injected in the main injection, main injection timing, inlet pressure, and engine speed. The two following parameters were varied, as shown in Table 3: -

Post injection timing. Indicated Mean Effective Pressure increase due to post injection.

Table 3: Table of sweeps. Post Injection Timing in CAD ATDC 12 14 15 16 18

IMEP Increase due to Post Injection in % 7 14 21 7 7 14 21 7 7 14 21

To analyze the transport velocity of soot under combustion with post injection, Particle Image Velocimetry (PIV) analysis was employed using combustion images taken by the HSV camera. In this work, no seeding technique was used to introduce tracer particles in the combustion chamber. Instead, the soot particles were considered as natural tracer particles. Being small and light enough, soot particles suit the role of tracers by following well the flow motion in the cylinder. The PIV calculations are based on the comparison of gray-levels between the images. The bright areas are due to the radiation of soot particles Figure 38.

Figure 38: Soot radiation at 20 and 21 CAD ATDC

60 A sequential cross-correlation method using normalized correlation function was used for all calculations. It compares the images in the sequence as follow: image 1 with image 2, image 2 with image 3 and so on… A good trade-off between dynamic range and spatial resolution was found using a double pass with decreasingly smaller interrogation windows size. The first pass was executed with 32 x 32 pixels interrogation windows and the second pass with 16 x 16 pixels. An overlapping of 50% was selected in order to obtain a better spatial resolution of the resulting velocity vector field considering the fact that one vector is generated by final window. As a result of this choice a final grid spacing of 8 pixels was obtained. The soot transport velocity was studied at four different points in the combustion chamber due to their relative position to the injector nozzle as shown in Figure 39.

Figure 39: The four studied points.

Due to large cycle to cycle variations in the velocity data, a strong averaging process was necessary. The method employed features three stages. The first one was a spatial average between the velocity vector at the considered point and its eight surrounding neighbours. The second step was the averaging between ten different cycles and finally a time average over 5 CAD was calculated.

4.4 Results and discussion NOx EMISSIONS

NOx emissions for all measurements show values between 22 and 25 part per million, showing that the variations of NOx level between all the configurations are very limited and can be considered as insignificant, as shown in some previous studies [1, 2]. Then, it was observed that the general NOx level measured is low considering that common values are of order 500 to 1000 ppm relatively to engine design and

61 operating conditions [3]. This point confirms that a low nitrous oxides emissions level can be secured even with a post injection event during the combustion. The constant level of nitrous oxides can be explained by the fact that post injection occurs during the expansion phase of the cycle. At this time, the temperature in the combustion chamber is decreasing due to the expansion phenomenon, and when this temperature reaches a certain value it freezes the NO chemistry. SOOT EMISSIONS The soot emission measurements as a function of post injection timing are plotted in Figure 40 it appears that 50% of the points are located under the reference line and are all located in the area plus or minus 40% soot emissions from the reference level. All the values obtained for a timing of 12 CAD ATDC exceed the FSN value of reference. A later post injection event seems to be effective in reducing soot emissions. However, a relative stabilization is observed between 14 and 18 CAD ATDC compared to a steeper reduction between 12 and 15 CAD. Soot (FSN) 1

% IMEP increase

0,9

7%

0,8

14%

0,7

21%

0,6

Reference value without post-injection

0,5 0,4 0,3 0,2 0,1 0 10

12

14 16 Post-injection timing [CAD]

18

Figure 40: Soot emissions vs post injection timing.

Soot emissions as shown in Figure 41 have a general tendency to increase with the increasing in IMEP which is related to the amount of fuel injected. However, more fuel means longer injection durations and local rich zones are created forming more soot. But the most noticeable point of this figure is that it shows that the injection timing can clearly be seen as an important factor. Post injection at 12 CAD ATDC is certainly too close to the end of the delay period and injects fuel in some already rich areas. The timings 15 and 18 CAD ATDC show a very similar behavior and even very close values.

62

Soot (FSN) 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 5

Post-inj. Timing 12 CAD ATDC 15 CAD ATDC 18 CAD ATDC Reference value without post-injection

10

15 % increase in IMEP

20

Figure 41: Soot emissions vs IMEP increase

SOOT TRANSPORT VELOCITY Very large cycle to cycle variations in the soot transport velocity were observed. As a result of these variations, single cycles could not be used as accurate data to compare the different post injection settings. Therefore a spatial and temporal averaging technique was applied for the whole velocity analysis. Figure 42 shows the soot transport velocity evolution during the combustion event for different post injection settings at point A in the combustion chamber. It appears that there is no logic concerning influence of post injection timing or fuel quantity on soot velocity that can be extracted from these results. The relative position of the different curves appears too random to draw proper conclusions. 7

no post injection post 12 CAD 7% post 12 CAD 21 % post 15 CAD 7%

6

m/s

5 4 3 2 1 0 0

10

20

30 CAD

Figure 42 : Soot transport velocity at point A

40

50

63 Nevertheless, it can be noticed that in almost all cases, the curves representing measurements with post injection show higher velocities in the late cycle (50 CAD ATDC) while the reference curve declines towards zero.

4.5 Conclusions The investigations on post injection settings and emissions showed the existence of optimal parameters leading to significant emissions reduction. As the post injection parameters were varied, no changes in NOx emissions level are observed even compared to a reference level established without post injection. The post injection events occur during the expansion phase of the cycle, the lower combustion temperature at this time explains their non-effect on NOx emissions. It means that the low NOx emissions level assured by the low temperature combustion remains unchanged. Optimal post injection timing exists to significantly reduce soot emissions. The delay between main and post injection has to be not too long or short to produce the maximum effect on soot concentration without penalty on hydrocarbons. The smallest amount of fuel post injected gives both the lowest soot and HC emissions level. In this study, the best post injection timing was 15 CAD ATDC, which corresponds to 21 CAD after the start of the main injection and the best fuel amount setting corresponds to an increase of 7% of the IMEP. This configuration with optimal timing and optimal fuel amount for the post injection leads to a soot emissions level reduced by more than 40% compared to the case without post injection. Constant low NOx level and significant reduction in soot emissions obtained simultaneously illustrate a rupture of the NOx-soot trade-off for Diesel engines. In the same time, HC emissions are increased by 10% which is one of the best results obtained for all runs. Velocity fields representing the radiating soot movement were obtained from the PIV calculations. Large cycle to cycle variations in the velocity fields led to the calculation of average vector fields. The averaging method included spatial, temporal and also multi-cycles average. The local velocity of soot particles was investigated from this averaged velocity data. The results obtained from the analysis of the local velocity of the transport velocity of radiating soot do not permit to conclude on the influence of post injection on higher turbulence and better fuel-air mixing. Therefore, enhanced late cycle oxidation due to increased temperature seems to be the key to soot emissions reduction.

4.6 Reference [1] [2] [3]

J.Benajes et al., “Influence of Pre- and Post-Injection on the Performance and Pollutant Emissions in a HD Diesel Engine”. SAE paper 2001-01-0526. F Payri et al., “ Influence of the Post-Injection Pattern on Performance, Soot and Nox Emissions in a HD Diesel Engine.” SAE paper 2002-01-0502. John B. Heywood: “Internal Combustion Engine Fundamentals”, McGrawHill, Inc. 1988

64

5. Study of Flame Lift-Off Length in Relation with Engine-out Soot in Heavy Duty Diesel Engine. Ongoing subproject. PhD students: Clément Cartier and Ulf Aronsson Flame lift off length is a parameter of interest when studying the origin of soot particles formation. Indeed, several studies [1,2] focused on this phenomenon in order to gain knowledge on the formation of pollutants. However, most of these previous studies were performed in combustion vessels where more favourable observation conditions for lift off observation can be achieved. Cycle to cycle variations, higher temperature and pressure gradients together with a variation of the combustion chamber volume are parameters than affects the accuracy of the measurements but can give a better picture of the mechanisms occurring in a real engine. A common definition of the flame lift off length (FLOL) is the distance between the nozzle and the beginning of the diffusion flame mantle as shown in Figure 43.

FLOL

Figure 43: Flame Lift off visualization on Dec’s model for Diesel combustion

It has been shown that a longer lift off length leads to lower emission of soot particles. This trend is illustrated in Figure 44 and can be explained by the fact that more oxygen can penetrate into the spray before the diffusion flame begins leading to leaner and more premixed combustion. This makes the combustion occurring in equivalence ratio - temperature regions where soot formation is significantly lower.

65

Figure 44: Relative soot incandescence versus Lift-Off Length [1]

Different parameters such as injection pressure and oxygen concentration in the inlet were varied in order to obtain an interesting range of FLOL and soot emissions. OHChemiluminescence was used in order to record images of the combustion and then obtain the FLOL.

Figure 45: Single picture of OH chemiluminescence from Diesel combustion

A pass band filter centered on 310nm was used in order to avoid recording the strong radiations from soot and to obtain better information on the diffusive mantle position characterized by higher OH radical concentration. Figure 45 is a picture extracted from a high speed video sequence recorded during the study. Details of the method is given in section 7 below.

66 To extract the FLOL from these images, image processing is needed. This procedure is not finished yet and will continue during the spring. However, some features of the Lift Off behavior have been observed so far. Figure 46 represents the FLOL for each of the eight sprays of the injector during the same cycle. Some significant variations can be observed between the sprays with a better convergence in the late cycle.

Lift off length for 8 sprays single cycle - 8 holes injector

40

Lift off length [mm]

35 30 25 20 15 10 5 0 -5 -10

SOI

-5

0

5

10

15

20

EOI

25

30

CAD

Figure 46: FLOL variation for each spray of an injector during a single cycle.

The end of injection is also noticeable by a fast and brief reduction in the FLOL followed by steep increase in the distance between the nozzle and the flame which propagates in a centrifugal way afterwards. More results on this study will be published as soon as the evaluation process of the data is finished. REFERENCES [1] Dennis Siebers and Brian Higgins, “Flame Lift Off on Direct-Injection Diesel Sprays Under Quiescent Conditions”, SAE paper 2001-01-0530 [2] Brian Higgins and Dennis Siebers, “Measurement of the Flame Lift-Off Location on DI Diesel Sprays Using OH Chemiluminescence” , SAE paper 2001-01-0918

67

6. Comparison between optical and all metal engines 6.1 Personnel The engine was run by Ulf Aronsson, Clément Chartier and Leif Hildingsson.

6.2 Background A lot of experiments are performed on optical engines in order to understand the combustion process in ordinary engines. In spite of this little work is done to verify that the model is representative. The combustion chamber is constructed differently in optical and ordinary engines to obtain optical access. This analysis is performed to find out how these differences are interfering with the combustion process. The main factors that are compared are the speed of the heat release and the ignition delay.

6.3 Known differences The investigated optical engine configuration is a Volvo D5 engine of Bowditch design. This type of optical configuration is commonly used in engine experiments, see Figure 47.

Figure 47: Volvo D5 engine with a Bowditch design.

In the upper part of the cylinder liner a quartz ring is fitted in and the piston is extended so that it’s possible to place an angled mirror in the bottom. The combustion chamber is then visible from below via the mirror, through a quartz window in the piston top. These modifications result in unavoidable differences between optical and ordinary engines. The walls of the combustion chamber feature different heat conduction characteristics because quartz is a much poorer heat conductor than steel or

68 aluminum. There is larger crevice volume in the optical engine since the piston rings has to be placed lower so that they don’t scratch the quartz window of the cylinder liner. There is also poorer cooling system in the optical engines because no water canals go through the quarts ring and no oil is cooling the surface under the piston.

6.4 Test Conditions Three different engine configurations are included in the test matrix. An optical engine of Bowditch design, one similar engine with all quartz parts replaced with metal ones and a standard diesel engine. These three engine configurations have the same shape regarding the combustion chamber and the same type of injection system. Two sweeps where the injection timing is changed and two sweeps where the inlet oxygen concentration is varied are chosen as running conditions. For all compared cases the duration, timing and pressure of the injection are the same. The in- cylinder condition, meaning pressure and temperature at top dead center are also held constant. Tests on the optical engine and on the same engine with metal parts have been performed so far, same tests are going to be performed at the standard diesel engine later on.

6.5 Results For all compared cases the combustion takes place earlier when the quartz parts are fitted in the engine. The result is quit obvious, even at running conditions for which the temperature and pressures were slightly higher by mistake at the start of injection for the metal engine configuration. The presented results below are from that particular sweep. Figure 6-2 is showing one representative running condition.

J

Heat Release

500 400 300 200 100 0 -100

Optic Metal

-10

0

10 CAD

20

30

Figure 48: Heat release SOI=8 CAD BTDC, 15 % O2 inlet, T inlet =80 °C, P inlet =1.1 bar.

Figure 49 is showing the ignition delay for all running condition in one of the timing sweeps.

69

Ignition Delay 8.5 CAD

8 Optic Metal

7.5 7 6.5 9

8

7

6

5

4

Timing [CAD BTDC]

Figure 49: Ignition delay

In Figure 50 the crank angle position where half of the total energy is released are shown and Figure 51 is showing the combustion efficiency for the same timing sweep. As one can see the combustion efficiency is higher when there are optic parts inserted in the engine.

CAD ATDC

CA50

25 20 15 10 5 0

Optic Metal STD

9

8

7

6

5

4

Timing [CAD BTDC]

Figure 50: Crank angel degree for 50% burned. Combustion Efficiency 1 0.98 0.96

Optic Metal

0.94 0.92 0.9 9

8

7

6

5

4

Timing [CAD BTDC]

Figure 51: Combustion efficiency.

In Figure 52 two cases with different SOI are compared. In the metal configuration SOI is at 7 CAD BTDC and in the optic configuration SOI is at 8 CAD BTDC. The graph in figure 6-6 shows that the combustion isn’t necessarily faster with optical parts in the engine even though it takes place earlier.

70

J

Heat Release CA50 Const 500 400 300 200 100 0 -100

Optic Metal

-10

0

10 CAD

20

30

Figure 52: Heat release for two cases with different SOI.

6.6 Ongoing and future work Same tests that have been performed at the optical engine with quartz and metal parts are going to be performed on a standard diesel engine later this spring. Further analysis of the measurement data will hopefully give an answer to why the combustion takes place earlier when there are quartz parts in the engine instead of metal parts.

7. High-speed video campaign for measurements of simultaneous OH and natural luminescence and temperatures measurements with Two-color pyrometry. Staff Staff i.e. senior persons and Ph.D. students involved: Mattias Richter Ph.D., Robert Collin Ph.D., Johannes Lindén Ph.D. Student.

7.1 Introduction During 2006 measurements were performed employing the Phantom high-speed video system to the truck-sized Scania engine. The aim was to perform cycle-resolved, lineof-sight diagnostics of combination in a parametric study where the engine was running at different operating conditions varying the injection pressure, the boost pressure, and the size of injection nozzle. A secondary aim was to investigate the potential of the high-speed video system for advanced engine investigations by applying different measurement approaches. Recordings of OH chemiluminescence and also recordings of OH luminescence combined with simultaneous detection of natural luminescence using a stereoscope were acquired with the system. Another investigation was focused on spectral investigation of the cycle-resolved natural luminescence using the high-speed video system combined with a spectrometer and thereby evaluate the feasibility of using a band-pass filter for OH detection. Finally, cycle-resolved high-speed video recordings of two-color pyrometry were performed in order to deduce temperature information from the burning fuel sprays.

71

7.2 OH luminescence and simultaneous OH and natural luminescence using a stereoscope Table 4 summarizes the different measurement conditions with varying injection pressure that were executed for the fixed boost pressure of 2 bar. For the simultaneous recordings using a stereoscope, one channel was equipped with filter for OH luminescence and the other channel was recording the un-filtered natural luminosity. Table 4: Measurement cases for the high-speed video recordings using OH filter and the recording using stereoscope, with OH band-pass filter (310nm) and natural luminosity in different channels. Boost pressure 2 bar 2 bar OH+Nat.L.* OH OH+Nat.L. OH OH+Nat.L. OH Small & large nozzle Small nozzle 9 kHz 36 kHz (1 spray only) Small nozzle: 0.10 mm, large nozzle: 0.14 mm * measured using only small injection nozzle Inj. Pressur

Engine operating condition 1400 bar 1700 bar 2000 bar Nozzle: Frame rate:

Figure 53 shows an example frame from a high-speed video recording of the simultaneous OH (310 nm band-pass) and natural luminescence signals. The injection pressure for this case was 2000 bar, a small injection nozzle was used, and the boost pressure was 2 bar.

Figure 53. One single frame from of a high-speed video recording of simultaneous OH (band-pass filtered) to the left and natural luminescence (to the right) using a stereoscope.

The operating condition in Figure 53 is a sooty case and since the heat radiation from soot has a broad spectral distribution the signal detected through the OH filter does not necessary have to originate only from OH but may also originate from the soot radiation. By using the stereoscope to visualize the signal detected through the OH filter and comparing it to the signal detected through the channel without any filters

72 (which signal is dominated by the radiation from soot) makes it easy to distinguish what part of the OH signal that is originating from soot radiation and what part is originating from OH luminosity. As an example, the “legs” close to the nozzle probably originate from OH luminosity. This observation is based on the comparison of the OH frame with the reference image to the right containing the detected natural luminosity where these “legs” are not that pronounced. By comparing the OH signal with the reference signal, exemplified in the frame in Figure 53, it can be interpreted that OH seems to surround the sooty parts as a shell. It should be noted however, that the image discussed is detected using a line-of-sight technique and is not a 2D measurement using a laser diagnostic technique which makes the interpretation less certain. In Figure 54, two selected frames from different spectra recorded with the high-speed video system combined with a spectrometer are shown. On top a spectrum recorded without any filters is shown, and below, the spectrally resolved signal for the same engine condition is shown, but with the OH filter applied.

Figure 54: Two selected spectra obtained during combustion. On top, spectra obtained without the use of any filters, and below, spectra obtained applying the OH filter.

The two spectra, showed in Figure 54, are both recorded during combustion and with the same engine operating condition. The spectrum on top is recorded without the use of filters and is thus mostly showing signal from the heat radiation from hot soot. For the shortest wavelengths presented (