Gas Engine Project

1
Gas Engine Project

• Mehrzad Kaiadi
• Per Tunestål
• Bengt Johansson

2
Outline

• Project history
• Achieved Results from the Previous Work
• Present Project
• Test-Cell Status
• Initial Measurements
• Follow-On Program / Schedule Suggestions

3
Project History

• 1990Flame propagation and turbulence
• 1995Combustion chamber effect on turbulence,
  combustion and emissions(single-cylinder engine
  Volvo TD102)
• 1997-2003Experiments on the multi-cylinder
  engine (6 cyl. Volvo TG103)

Goals Improve Engine Stability, Efficiency
and Output Power Reduce Emissions
4
Achieved Results from previous Work

• Combustion chamber effect on turbulence and
  combustion
• Lean operation versus Stoichiometric operation
  diluted with EGR

5
Achieved results (Turbulence)
Reentrant Square
Reentrant Cylindric
Fair Top
Nebula
Quartette
Turbine
6
Achieved results (Turbulence)
In-cylinder flow and heat release
7
Conclusions of the previous work

• Stoichiometric operation diluted with EGR is
  recommended
• Highly diluted mixtures have cold and slow
  combustion
• Quartette combustion chamber was designed to
  generate high turbulence and speed up combustion

8
Present Project

• Will be based on the previous results New Ideas
  
• Stoichiometric operation diluted with EGR

9
Test Engine (Volvo 9.4 Liter)
10
New Features

• Multi Port fuel injection is designed (Single
  point injection will be replaced by this)
• Advantages
• Improved volumetric efficiency
• More uniform fuel distribution
• Rapid engine response to change throttle position
• Better control of lambda during cold-start
• Results in increasing power and torque

11
New Features
12
New Features

• New Ignition and Injection System is developed
• Cylinder-individual control of fuel injection
• Ion-current signals (will be available in near
  future)

13
Established Engine Test-Cell
14
Established Engine Test-rig
15
Engine Control System
16
Engine Control System (Interface)
17
Running Method

• Stoichiometric operation diluted with EGR
• Impelementing HEDGE concept In order to achieve
  diesel engines efficiency and otto engines
  emissions level (Use high EGR rates coupled with
  high energy ignition system)

18
Why HEDGE ???

• Diesel Engines
• High efficiency
• High emissions (Soot NOx)
• - Catalyst (Costly reduce efficiency)

• Gasoline Engines
• Reasonable emissions (By running
• Stoichiometric using 3way catalyst)
• Lower performance
• - Lower fuel economy

How to get diesel efficiency ?
High Efficiency diluted Gasoline Engine (Good
candidate) Reasonable emissions Air-flow boost
will increase the performance Highly diluted
with EGR can increase the knock limitations and
decrease NOx
19
HEDGE Structure
20
Initial Measurements

• Finding out
• lean limits (lambda EGR)
• Behavior of the engine (cylinder variation)
• EGR Vs. Lean

21
Initial Measurements

• 1200 RPM
• Constant air flow / variation of fuel flow

2 Reasons Big cylinder to cylinder variations
Not optimal combustion chamber (no turbulance)
22
Cylinder to Cylinder variation
23
Cylinder to Cylinder variation
24
Initial Results (Lean/EGR)

• Raw hydrocarbons, HC

Lean operation (?gt1)
EGR operation (?1)
25
Initial Results (Lean/EGR)

• Raw nitric oxides, NOX

Lean operation (?gt1)
EGR operation (?1)
26
Initial Results (Lean/EGR)

• Raw carbon monoxide, CO

Lean operation (?gt1)
EGR operation (?1)
27
Brake Efficiency
28
Learning from the first running

• New intake manifold should be designed
• Engine behaves in expected manner

29
Follow-on Program / Suggestion

• Modeling
• GT-Power Model of the engine
• Running Miller cycle
• Artificial Neural Networks (ANNs) Modeling
• Experiment
• Studying Closed-loop combustion control to
  optimize the amount of fuel and EGR (by
  controlling lambda and COVimep)
• Studying the effect of cylinder variation
• Transient runnings
• Implementing ion current measurments

30
GT-Power Modeling

• Investigating Higher Performance Potentials
• SI engines have poor performance at part and
  light loads
• Miller cycle can be a solution
• VVT
• Supercharger
• since it needs extra components modeling is the
  most cost-effective way

31
GT-Power Modeling

• The expansion ratio is greater than compression
• By allowing inlet valve be open for a longer time
• gt effective comprestion ratio decreases
• gt Better efficiency
• Excepted Results Higher efficiency at low loads

32
Closed-Loop Combustion Control

• Lambda Control For controlling the overall
  air/fuel ratio
• COVimep Control Since the engine is run with
  high rate of EGR, the variation in mixture
  composition especially near the spark plug can be
  high and thereby the operation will be rough and
  unstable

33
Closed-Loop Combustion Control

• Expected results Excellent steady-state
  performance and limited transient performance

34
Studying the Effect of cylinder variation

• Why?
• Better understanding of cycle variations effect,
  and how to control them
• How?
• By changing ignition timing or / and lambda
  (injection offset) of only one cylinder find out
  how the other cylinders behave

35
Projects Time Schedule
36
ANNs Modeling (Suggestion)

• What is a Neural Network?
• Definition An Artificial Neural Network (ANN) is
  a computer program that can recognize patterns in
  a given collection of data and produce a model
  for that. It resembles the brain in two ways
• Learning from experience (trial and error)
• Interneuron connections known as weights are used
  to store the knowledge

37
ANNs Structure

• Artificial Neuron

• An ANNs is a set of processing elements and
  connections with adjustable weights

Inputs
Outputs
38
How dose an ANN work?

• Train the Network
• Present data into networks
• Network computes an output
• Network output compared to desired one
• Networks weights are modified to reduce the error
• Generalization to new data
• Use the Network
• Present new data into networks
• Networks compute an output based on its training

39
ANNs Applications

• Sensitivity analysis
• Tracking degradation
• Making a user-friendly interface of the engine in
  e.g. Excel
• Online-Monitoring / Using in control loops

40
Supervising ANN (online-monitoring)

• Early varning system

41

• Thanks for Your Attention!
• Questions?