Model Based Control Algorithm for a Bi-Fuel Engine with fuel adjustment

(1)

Model Based Control Algorithm for a Bi-Fuel Engine with fuel adjustment

JOHN ERIKSSON

MÅRTEN WALLENGREN

(2)

.

Examensarbete MMK 2013:91 MDA 468

Modellbaserad controll algoritm fr en Dual-brnslemotor med brnsleanpassning

John Eriksson M˚ arten Wallengren

Godk¨ ant: Examinator: Handledare:

2013-12-17 Mats Hanson Bengt Eriksson

Uppdragsgivare: Kontaktperson:

AVL M˚ arten Wallengren

.

Sammanfattning

.

F¨ or lite mer ¨ an 100 ˚ ar sedan rullade den f¨ orsta serietillverkade bilen, i den helt valbara f¨ argen svart, av monteringslinan. Sedan dess har ett febrilt utveckingsarbete gjorts f¨ or att f¨ orb¨ attra alla delar av bilen. F¨ orst endast i marknadsf¨ oringssyfte med st¨ orre motorer och mer komfort, men de senaste decennierna ¨ aven som svar p˚ a ut¨ okade lagkrav runt om i v¨ arlden for att f˚ a s¨ alja p˚ a respektive marknader.

En av de metoder som p˚ a senare tiden har vuxit f¨ or att minska bilarnas utsl¨ app

¨

ar att g˚ a ¨ over ifr˚ an bensin och diesel till alternativa br¨ anslen. D˚ a finns det b˚ ade f¨ ornyelsebara samt icke f¨ ornyelsebara, men med h¨ ogre verkningsgrad ¨ an konven- tionella br¨ anslen, att v¨ alja p˚ a.

F¨ or att st¨ odja denna utveckling s˚ a har Anstalt f¨ ur Verbrennungskraftmaschinen List (AVL) beslutat att forts¨ atta utveckla sin motorstyrvara, kallad Rap2L f¨ or att

¨

aven kunna styra motorer drivna av Compressed Natural Gas (Komprimerad natur- gas) (CNG).

Detta arbete syftar till att utveckla en modul¨ ar gasfunktionalitet som b˚ ade ska

kunna anv¨ andas f¨ or att k¨ ora Rap2L, men ¨ aven enkelt kunna flyttas ¨ over till ett

annat motorstyrsystem om ¨ onskan f¨ or det skulle uppkomma.

(3)

.

Examensarbete MMK 2013:91 MDA 468 Model Based Control Algorithm for a Bi-Fuel Engine

with fuel adjustment John Eriksson M˚ arten Wallengren

Godk¨ ant: Examinator: Handledare:

2013-12-17 Mats Hanson Bengt Eriksson

Uppdragsgivare: Kontaktperson:

AVL M˚ arten Wallengren

.

Abstract

.

A little more than a hundred years ago the first mass produced car rolled of the as- sembly line, in the optional color black, and since then a hectic race has been going on in order to improve all parts of the car. First only as a competitive edge with bigger engines and more comfort for the passengers, but in the last decades also as a response to increased legislation’s around the world in order to be able to sell to their respective market.

Apart from increasing the effectiveness of both engine and catalyst in order to meet the ever increasing environmental demands, there is also the possibility of exchanging the fuel used. One prominent alternative here is Compressed Natural Gas (CNG), that can be extracted both from natural gas or biogas, dependent on what available resources are at hand.

In order to accommodate this change in fuel Anstalt f¨ ur Verbrennungskraftmaschinen List (AVL) has the desire to further develop its engine control software Rap2L, in order to also be able to run CNG engines.

The purpose for this thesis is to build a separate add-on to Rap2L where all the

functionality for CNG is handled, it should also be able to, with minimal effort, be

transported to another engine control system, if so desired.

(4)

(5)

Nomenclature

Abbreviations

Symbol Description

AFA Alternative Fuel Adjustment

AFR Air-to-Fuel ratio

AVL Anstalt f¨ ur Verbrennungskraftmaschinen List

BDC Bottom Dead Center

CI Compression ignited

CNG Compressed Natural Gas

CO Carbon monoxide

ECM Engine Control Module

ECU Engine Control Unit

EMS Engine Management System

HC Hydrocarbon

HFM Hot film airflow meter ICE Internal Combustion Engine LPG Liquefied Petroleum Gas

MON Motor Octane Number

MVEM Mean Value Engine Model NEDC New European Driving Cycle

NO

_x

Nitrogen oxides

(6)

OEM Original Equipment Manufacturer

Pm Particle matter

RON Research Octane Number

SI Spark ignited

SPI Single Point Injection

TDC Top Dead Center

(7)

Important definitions

Knock As the flame propagation velocity of the air-fuel mixture is significantly lower than the speed of sound in the mixture, the unburnt part of the air-fuel mixture will experience a sharp pressure increase as the fuel burns.[10] This pressure increase can cause the fuel to auto ignite explosively, causing large vibrations in the engine and significantly decreasing the lifespan of the engine if allowed to continue.

Octane number The tendency for knocking to occur in a particular fuel. Higher number means a lower tendency for knocking. Often described in Research Octane Number (RON) or Motor Octane Number (MON). RON is often used for simulating the conditions at lower loads and speeds while MON is used for higher loads and speeds.

Cetane number The tendency for the fuel to auto-ignite when injected into a Diesel type engine. Higher Cetane number means a shorter ignition delay and there- fore smoother combustion and quieter engine.

Stoichiometry Stoichiometry is the science of measuring different substances.

When the reactants of a chemical reaction is balanced in order to react without any residuals, it is called that the reaction are stoichiometric.

Lambda Lambda is the ratio between the stoichiometric Air to Fuel-ratio and the actual Air to Fuel-ratio in the combustion chamber. When lambda is larger than 1, with exes air, it is called that the engine is using a lean mixture, while if lambda is less than 1 it is called a rich mixture. [11]

Well-to-wheel A concept used to take into account the entire cycle of the fuel, all the way from the source to useful energy at the wheels.[8]

Mean Value Engine Model (MVEM) Model of a engine where cycle-to-cycle variations are ignored and the cylinders are modeled as single combustion chamber where a continuous combustion occurs. This is done in order to significantly speed up the computations while still giving satisfactory results from the model.

Gasoline normalized In order to be able to assess the potential gains of different

fuels the different parameters like energy consumption and emissions are normalized

with gasoline.

(8)

1 Introduction

Where a short introduction of the company is given.

AVL was founded in 1948 by Dr. Hans List and is today one of the major consulting firms in the development of Internal Combustion Engines (ICE), electric power-train systems as well as instrument for, and testing of, such systems.

When developing a new engine, and accompanying Engine Control Unit (ECU), the time for calibration can, due to the large numbers of parameters and repetitions needed, easily run in the excess of 12 months. In order to reduce this time, AVL has developed AVL RAP2L, where both the on- and off-line simulations run on the same controller. This approach enables the controller to be made in parallel with the en- gine, and there by make it possible to find many of the bugs early in the process and eliminating many of the errors that can occur when porting between the simulation and the actual ICE.

In order to run the controller off-line before the engine is made, a complete model- in-the-loop environment has been developed, where a complete vehicle model can be inserted in order to simulate the dynamics off the engine and car. A MVEM is used to simulate the combustion as it gives a good trade off between accurate data and a fast simulation. As all cylinders are not modeled, it can not give accurate predic- tions of dynamics on the time scale of a few cylinder revolutions. But over greater time-spans crankshaft speed and manifold pressure, as well as internal variables such as volumetric and thermal efficiency’s, give good results.

In recent years the demand for cars running on alternative fuels has drastically in-

creased, as the evidence of human impact on our planet has become more and more

evident. The purpose for this thesis is to expand Rap2L with functionality to be

able to run on alternative fuels, specifically a mix of gasoline and CNG.

(11)

2 Background

The different aspects when converting diesel and gasoline engines to run on CNG, the emission benefits of CNG, the idea with Rap2L and the rules concerning retrofitting a CNG system to a car versus building a bi-fuel car are discussed.

AVL is one of the major private supplier of engineering and development resources in the world, servicing all parts of the automotive industry. One of the services AVL provides to the customers is rapid prototyping of different systems in the whole ve- hicle. For engine development there are several areas which have to be adapted or optimized for CNG and therefore there is a great advantage of rapid prototyping.

Rap2L is a rapid prototyping tool replacing the standard engine management system.

By replacing the whole Engine Management System (EMS) with a rapid prototyping system, which is completely Simulink based, combined with a highly flexible hard- ware, changes to the system could be tested very quickly. New hardware or sensors could be added to the engine and a first algorithm to control it can then be made in Simulink and tested immediately without going through the whole process of order- ing a software and hardware change from the EMS supplier. This means part of the system could quickly be replaced with a new one provided by rapid prototyping and the feasibility of the new part can be tested and verified in a very short time frame, thus shortening the very important time to market.

Alternative fuels have a growing importance for all customers all over the world and one of the most important are gaseous fuels like CNG and biogas. The use of CNG in automotive applications is realized by mainly two engine technologies; Otto engines and dual fuel in diesel engines. The dual fuel technology uses the diesel engine as base and replaces some of the diesel fuel by CNG while still working ac- cording to the diesel cycle using diesel combustion to ignite the CNG air mixture.

The combustion becomes complex, mixing diffusion and flame propagation combus- tion. Combustion of CNG with Otto engine technology on the other hand has no major differences depending on the fuel; the air fuel mixture is ignited with a spark plug and combustion occurs via flame propagation.

Both technologies suffer from the properties of methane as fuel, methane being the

dominant type of CNG, making it necessary to adopt the control strategies of the

engine when running on CNG. Methane is a quite stable molecule making it more

difficult to combust compared to the hydrocarbons found in gasoline or diesel, but

(12)

when completely burned CNG fuel results in allover lower emissions. For example both CO and CO

₂

emissions will be about 20% lower compared to gasoline due to the fact the methane molecule contains less carbon than the major types of gasoline.[8]

In the automotive market both Dual fuel and Otto engine applications exists. Dual fuel systems are up till now mainly implemented as aftermarket systems only ap- proved locally, whereas Otto engines for CNG are designed and sold by major man- ufacturers. Otto engines are also the technology which is verified to meet all legal emission standards over the world.

The Rap2L system is mainly designed to work with light-duty engines for car ap- plications that run according to the Otto cycle. Being as flexible as it is, it could of course also be modified for other applications as heavy duty and with Dual fuel technology, but then major changes needs to be implemented. In car applications only Otto engine technology is used with CNG as it is more practical and less costly to modify an existing Otto engine car. When taking into account that the legal emissions and diagnostic standards have to be met with CNG, all this makes the Otto engine alternative the only viable solution.

Since CNG fuel is not available everywhere most cars are made as Bi-fuel appli- cations, meaning that they could be run on both CNG and gasoline. The common way of making a Bi-fuel car is to start from a standard gasoline car and then altering only the parts needed to also run on CNG, and the EMS system being one of the most important parts to modify.

This thesis work was aimed at making changes to the existing Rap2L system to convert it to a Bi-fuel engine management system. The considerations which had to be taken where that the CNG changes should be kept isolated from the rest of the system making it more of a plug-in application, than a completely new software and keeping the architecture as generic as possible. The goal is that the same require- ments that has to be met for gasoline (e.g. drivability, emissions) should also be met when running on CNG.

This idea of making a plug-in application for Bi-fuel could potentially be something

that AVL’s customers could be interested in. Keeping their own gasoline system

which is verified through all their development process intact with calibration data

and all, and only making small changes to implement a package making the whole

system into a Bi-fuel software.

(13)

Historically there have been mainly two kinds of Bi-Fuel cars on the market; whole car type approved vehicles and retrofitted cars. The whole car type approved vehicles are cars which were developed fully by the Original equipment manufacturer (OEM) and approved as a Bi-fuel car according to the same regulations as a normal gasoline car. However, the most common type is the retrofitted Bi-fuel car. This is where the OEM isn’t doing the design and installation of the CNG system, it is instead made by a retrofit company who has specialized themselves in doing aftermarket CNG kits for many different cars.

The retrofit systems are similar in design but the level of quality and functional complexity varies. When retrofitting a car with a CNG system only two special reg- ulations has to be fulfilled; r115 and r110, as long as the car is already type approved in EU as a gasoline car.

The control of the standard retrofit system is normally done by a so called piggy-back ECU. This means that the new ECU sits separately on top of the gasoline standard ECU and works in a way that signals in and out of the gasoline ECU is read and manipulated to make the gasoline ECU work with CNG aswell. This design imposes limitations on the quality and of drivability of the car. The control of the engine when running on CNG is also affected, in worst case, causing damage to the engine and the catalytic converter.

The OEM’s want to avoid that their cars, which are retrofitted with CNG sys- tems, have quality or drivability issues and are therefore in some cases working with the retrofit company, enabling some communication between the gasoline and the CNG ECU. By doing so the quality could be raised and some of the issues could be avoided. However the r115 regulation does not allow the OEM to do the acctual retrofitting, it has to be done by another company.

A next step of retrofitting would be if the OEM allowed the retrofitting company to add CNG functionality directly in the gasoline ECU. Then in theory all quality and drivability issues caused by the CNG systems could be avoided. This solution could at first glance seem to be the perfect one, but it is not so easy to do it in reality.

The OEM’s are responsible for the gasoline ECU and the software in it, if a retrofit

company alter the software, who will then take responsibility for the complete pack-

age? The OEM will probably not do it because the CNG software will not be verified

and validated according to their methods and software development process. The

(14)

small retrofit company can also not do it because they cannot take responsibility

for issues which could be caused by the standard gasoline software. Of course these

kinds of problems could be handled by thorough agreements between the OEM and

the retrofit company done prior to introducing the car. The potential benefits of this

technical solution should be a driving force enough to realize such an agreement.

(15)

3 Purpose

Why are we doing this?

The idea with Rap2L is to be one such rapid prototyping system mentioned above, where the aim is to get a first working prototype of an engine up and running as quickly as possible. As such, very little weight is given to diagnostics and error de- tection in the general Rap2L, and none in this thesis.

The purpose with this thesis is to investigate, and implement, the necessary changes to AVL’s internal rapid prototyping engine control system Rap2L. The changes to this program should be made in a as modular way as possible in order to, with min- imal additional work, port the functionality to another control system.

In order to verify the software from this project some time shall also be given to investigate the possibility of implementing the code on a real engine.

3.1 Objectives

• Implement functionality for CNG in Rap2L.

• Enable the CNG-software to be portable to another engine controller.

• Investigate the possibility of implementing dSPACE hardware on a existing bi-fuel car for testing.

3.2 Delimitation’s

• No fault detection or correction

• No calibration

(16)

3.3 Thesis overview

An overview of who wrote what.

• Introduction.

Where a short introduction of the company is given.

– Author: John Eriksson

• Background.

The different aspects when converting diesel and gasoline engines to run on CNG, the emission benefits of CNG, the idea with Rap2L and the rules con- cerning retrofitting a CNG system to a car versus building a bi-fuel car are discussed.

– Author: M˚ arten Wallengren

• Purpose.

– Author: John Eriksson

• Frame of reference.

A brief overview of different fuels in use, as well as a short introduction to engines, then deeper into what is needed both from hardware and software to run on CNG. The chapter is ended with an part about Rap2L and how it is built.

– Author: John Eriksson.

ICE composition, ICE implementation and Rap2L overview – Author: M˚ arten Wallengren.

CNG functionality and CNG quality adaption.

• Method.

We dive deeper into Rap2L and the thoughts behind the changes done to it.

– Author: John Eriksson

• Results.

The results are given.

– Author: John Eriksson

(17)

4 Frame of Reference

A brief overview of different fuels in use, as well as a short introduction to engines.

A deeper view into what is needed both from hardware and software to run on CNG.

The chapter ends with an part about Rap2L and how it is built.

4.1 Overview of fuels

In order to put CNG in its context in the ever growing market of alternative fuels, today approximately 15 million vehicles, a short summary of different fuels for ICE is given. It is by no means exhaustive and is only meant as an overview in order to give a perspective of what is available.[1]

Table 1: Fuel characteristics [8] [4]

Energy content [KJ/Kg]

Octane number [RON]

Cetane number

white

Normalized well-to-wheel energy

white

Gasoline 46.5 90-95 n.a. 1

Diesel 45.7 n.a. 48-50 0.75

LPG 50.2 107.5 n.a. 0.83-1.02

Ethanol 29.8 109 8 1.17-2.69

Methanol 22.9 110 5 1.10-1.65

Hydrogen

¹

142.2 106 n.a. 1.78-3.46

CNG

¹

52.2 120 n.a. 0.88-0.91

1

At standard atmosphere and 0

^◦

C

4.1.1 Gasoline/Diesel

As the two fuels first industrialized they enjoy a huge advantage from the fact that infrastructure is already in place and the huge volumes makes it possible for the distributors to have a relatively low mark-up, helping in keeping the prizes down.

Both carry the disadvantage of not being biodegradable, giving them a big environ-

mental impact in case of a leakage, but they can be considered relatively safe, both

since they are in a stable form as well as the general public has learned to handle

(18)

them.

Both have relatively low well-to-wheel energy consumption, where Diesel has the low- est of all fuels here shown. Diesel has high emissions of Nitrogen oxides (NO

_x

) and Particle matter (Pm) and performs better than gasoline on Carbon monoxide (CO) and Hydrocarbon (HC) while Gasoline has lower NO

_x

and higher CO emissions.[8]

4.1.2 Liquefied Petroleum Gas

Liquefied Petroleum Gas (LPG) (Propane/Butane) is harvested at the same time as other petroleum based fuels. It is gaseous at ambient temperatures and pressures and is compressed and stored as a liquid at 6-8 bar. Due to it’s lower energy content it needs twice the volume and 1.5 times the weight of Gasoline for the equivalent energy.

As LPG is heavier than air it will tend to linger at the location if a leakage and settle in low-spots. For this reason it is sometimes prohibited for LPG-vehicles to use underground parking and a strong aromatic is added to give it a distinctive smell.

The well-to-wheel energy consumption of LPG is higher than that of Diesel but lower than Gasoline. [8]

4.1.3 Ethanol/Methanol

Ethanol and methanol are both alcoholic liquids with similar properties. They both have a lower energy density than Gasoline and are more expensive to manufacture.

Ethanol predominately comes from biomass while methanol can come from both biomass or petroleum sources, but it is today to expensive to extract it from biomass in significant volumes.

Both are predominately used in combination with Gasoline, where they help lower the well-to-wheel emissions of NO

_x

and can be used as a anti knock additive due to its higher octane number.

The great difference in well-to-wheel energy for ethanol comes from the different ways of production. Ethanol from cellulose have a gasoline-normalized well-to-wheel consumption of 1.78-2.69, and from sugar/starch it is 1.17-1.51.

Both ethanol and methanol are biodegradable and their evaporation rates are lower

than gasoline, leading to lower risks in the event of a accident. However, their air-fuel

mixture ranges where explosive combustion can occur are larger, so in the event of

an accident in a closed environment the danger of explosion is increased.[8][11]

(19)

4.1.4 Hydrogen

Hydrogen is the lightest element in the periodic table and can come from a number of different sources with very varying impact on the environment, from electrolysis, reforming natural gas or gasification of coal/biomass. As the combustion in a engine gives, apart from NO

_x

, negligible emissions, the environmental impact is heavily influenced on the production method used, this is also the source of the great disparity in the normalized well-to-wheel analysis.

The energy content of hydrogen per mass is high, but by volume it is very low, giving the necessity of very big tanks. If hydrogen is stored at 200 bar it requires 20 times the volume compared to gasoline, for the same energy content.

Another option is to chemically bind the hydrogen to a metallic substance using a process called hydride, this gives an acceptable volume, but the weight then becomes 10 times that of gasoline.

The third option is to use liquefied hydrogen. However, this requires isolated tanks able too keep the hydrogen at -253

^◦

C, which gives a weight of 1.5 times and a volume at 4 times that of gasoline.[8][11]

4.1.5 CNG

Typically consists of about 80-98% methane and can be used in typical Otto/diesel- engines with only minor hardware adjustments. It has both a higher energy content and octane number than gasoline. However, due to the fact that the gaseous form takes up more volume in the combustion chamber, the total energy output at full load is lowered by approximately 10% compared to gasoline.

Both the well-to-wheel energy and well-to-wheel emissions is lower than that of gaso- line and diesel.

One of the major hazards with using CNG in automotive applications is to avoid a

leak when an accident occur. If a full load of CNG would be allowed to leak out into

the air around an accident, it would create a far larger flammable volume than the

equivalent gasoline quantity. On the other hand the gases is easily carried away by

wind.[8]

(20)

4.2 ICE implementation

4.2.1 2-stroke engine

A 2-stroke engine derives it name from the fact that it completes its cycle in just two strokes, or one full rotation of the crank shaft. The cycle starts at Top Dead Center (TDC), when the spark plug discharges and ignites the air-fuel mixture. This induces an explosive combustion in the chamber, driving the piston downwards. When the piston gets close to Bottom Dead Center (BDC) it uncovers, first the exhaust port and then the intake port, located in the side of the cylinder walls. The piston is then forced upwards by the momentum of the flywheel to TDC and the cycle is complete.

Figure 1: 2-Stroke engine during compression. On the right side the exhaust port

is just about to be closed, while on the left the intake port has just been closed. In

orange is also shown the rotary valve used to open and close the intake port to the

pre-compression chamber, shown as a white ring in the back wall.

(21)

One great challenge with 2-stroke engines is to design the ports and combustion chamber in order to expel all of the exhaust while minimizing the amount of fresh gases that goes out into the exhaust.

In order to achieve a more homogeneous mixture and increase the speed by which it is flowing into the combustion chamber the air and fuel are first injected into the volume directly underneath the piston. This has the benefit of not only compressing the mixture prior to flowing into the combustion chamber by the downward mov- ing piston, but also that the constant flow of gasoline acts as a lubricant for the crankshaft and connecting rod.

4.2.2 4-stroke Otto engine

In the 4-stroke engine the compression, that occurs below the piston in a 2-stroke engine, has been moved into the actual combustion chamber making the cycle consist of 2 full revolutions.

Intake stroke Intake starts just before the piston is reaches TDC, and is composed of the first half of the first revolution down to BDC. During this period the downward motion of the piston is used to pull a fresh mixture of air and fuel into the combustion chamber.

Compression stroke Starts at BDC and goes back up to TDC as the piston compresses the mixture and ends with the spark plug firing some 0-30

^◦

before TDC to start the combustion. The precise angle of ignition is a trade off between a desire to have the maximum pressure as close to TDC as possible for increased efficiency, but still avoid knocking.

Combustion stroke The combustion stroke is where the power is generated by the combustion of the air/fuel-mixture.

Exhaust stroke As the exhaust port is opened the exhaust experiences a so called

blow down, as the pressure in the chamber is much higher than the pressure in the

exhaust system, the piston then pushes the rest of the exhaust out as it moves back

to TDC and the cycle is complete.

(22)

4.2.3 4-stroke diesel engine

The diesel engine uses exactly the same four strokes as the Otto engine, the differ- ence being in that the diesel engine only takes in, and starts to compress, pure air.

Then, when the reaction is supposed to start, the diesel is injected into the cylin- der where it auto-ignites, due to high compression. Therefore no spark plug is needed.

Another major difference is the fact that the diesel engine is run largely unthrot- tled and load control is done only through varying of the amount of fuel injected.

Also, as the diesel engine is not limited by knock as Otto engines they have the

possibility to use a higher compression ratio in order to get higher efficiency in the

combustion process.

(23)

4.3 ICE composition

An ICE derives its power from the explosive combustion of one of several different fuels, the most common being gasoline and diesel, inside its combustion chamber.

In this thesis the focus is the Otto engine, also known as Spark Ignited (SI) engine, so named after Nikolaus Otto who invented it in 1876. Otto engines runs on a com- pression ratio of about 8-12:1 and uses a spark plug to ignite the air/fuel mixture, as opposed to Diesel engines, formally known as a Compression ignition (CI) engine, that uses a compression ratio of 15-20:1 and let’s the fuel auto-ignite as it is injected into the combustion chamber.[7]

Figure 2: A schematic overview of an engine.

A, Intake manifold, with throttle. B, Combustion chamber. C, Exhaust manifold

(24)

4.3.1 Air intake

Air filter The air first passes through an air filter on its way into the engine in order to filter away large particles that might be harmful to the engine.

Hot film airflow meter (HFM) There are several versions of how to measure the mass flow of air that enters the engine. One common way is a so called HFM.

A HFM works by placing a thin wire in the airflow through which a current is then passed. As the resistance of the wire will change depending on the temperature and since the cooling factor from the air is proportional to the mass flow, not volume flow, the mass flow can be calculated.

Throttle As the SI engine requires a very exact air-to-fuel ratio in order to work efficiently a throttle is placed in order to be able to restrict the airflow.

Intake manifold Connecting the throttle to the intake of the cylinders is the intake manifold. One key aspect when designing the manifold is to consider the harmonics in the air flow as they pass through the intake manifold. As the ports are periodically opened and closed, a pressure wave will be created. This wave will travel up the manifold where reflections occurs due to the geometry of the pipe.

If the manifold geometry is poorly designed, certain frequencies can give rise to a standing wave with a pressure minimum at the port opening, leading to a drastic decrease in the amount of air entering the cylinder. By very carefully designing the manifold it is of course possible to generate a presure maximum at the port when it opens, giving a small boost to the system.

Injectors When using a port injected engine the injectors is placed in the manifold,

as close to the intake as possible. This gives greater control over how much fuel are

actually entering the cylinder on each combustion, as opposed to the old carburetor

that injected fuel to all the cylinders simultaneously at the throttle plate, called

Single Point Injection (SPI). A port injected engine also minimizes the part of the

manifold that comes in contact with the fuel, thus minimizing the phenomenon of

wall wetting that occurs when the fuel condenses on the cool surface around it.

(25)

4.3.2 Combustion chamber

Calculating A/F-ratio The catalytic converters in use today only works on a satisfactory level within a very narrow band around lambda = 1, that is, when the air to fuel mass fraction is the same as the stochiometrical mass fraction, explained further down. This leads to that a very specific amount of fuel needs to be injected to match the air coming from the throttle. In order to find the amount of fuel to inject to achieve a stoichiometric mass fraction the balanced chemical equation needs to be stated.

Table 2: Molar masses of different substances[14]

Substance Molecular weight [g/Mole]

Carbon: C 12.01

Oxygen: O 16

Hydrogen: H 1.008

Methane: CH

₄

16.04 Oxygen: O

2

32.00 Carbon dioxide: CO

₂

44.01 Water: H

₂

O 18.02

CNG

The balanced chemical equation for when CNG, here equated with pure methane, combusts with oxygen is:

CH

₄

+ 2 (O

₂

) ⇒ CO

₂

+ 2 (H

₂

0) (1) and with the molar masses shown in Table 2, the stoichiometric Air-to-Fuel ratio (AFR) becomes

_CH^2O²

4

=

_16.042^2·32

= 3.99. that is, for every gram of methane burnt 3.99g of oxygen is needed. As the air has a composition of 23.2 mass-percent oxygen and 75.5 mass-percent Nitrogen [6] not present in the reaction, the actual mass fraction of air needed is 3.99 ·

_23.2¹⁰⁰

= 17.2.

So the stoichiometric AFR ratio of methane is 17.2.

Octane

For octane the balanced chemical equation looks like:

2 (C

₈

H

₁₈

) + 25 O

₂

⇒ 16 CO

₂

+ 18 H

₂

O (2)

(26)

Using the same methodology as above, this gives

_C^25O²

8H18

=

_2·114.22^25·32

= 3.50, and with 23.2 mass percent oxygen in air it becomes: 3.50 ·

_23.2¹⁰⁰

= 15.1. Due to the fact that gasoline does not contain pure octane, but instead a mixture of a wide variety of petroleum molecules, the AFR is lowered to 14.7.

Swirl, Tumble and Squish In order to make sure that the fuel and air mixes thoroughly and is as homogeneous as possible, a turbulence has to be induced in the airflow. There are three ways to do this, and often a combination of all three is used.[9]

Swirl

Swirl is when a rotation parallel to the length axis of the combustion chamber is in- duced. This is the easiest to accomplish as the round shape of the cylinder will tend to deflect the incoming air around it. In Figure 3a one of the intake ports is disabled to increase the amount of swirl, giving approximately 1 − 3 rotations per combustion.

Tumble

Tumble is when a rotation perpendicular to the length axis of the combustion cham- ber is induced. As the piston is moving up and the height of the chamber becomes significantly smaller than the width, the tumble breaks up into many small vortexes, and thus greatly increasing the turbulence. See Figure 3b

Squish

Squish is when the edges of the piston head comes towards the cylinder head at TDC forcing the mixture in towards the center. See Figure 4

These three are used in conjunction when designing intake ports and combustion chamber in order to create an as homogeneous mixture as possible for a faster and cleaner combustion, leading to a higher efficiency and cleaner exhaust.

4.3.3 Air exhaust

Exhaust manifold As the combusted fumes exits the combustion chamber they enter the exhaust manifold which directs the individual flows into a single pipe.

Three way catalytic converter A three way catalyst is a ceramic or metallic

container covered by a coating with extremely high surface area, this is in turn

covered in platinum and rhodium, helping the gases convert into less harmful gases.

(27)

(a) Swirl in a cylinder, one intake port dis- abled

(b) Combustion chamber during intake, with Tumble

Figure 3

Figure 4: Zoom of the combustion chamber at TDC, showing squish

(28)

4.4 CNG considerations

The differences between running an engine on gasoline or on CNG are few, but some have a great impact on the hardware of the engine and the catalyst and are therefore important, while others have impact only on the drivability of the car. The latter is not less important since it affects the experience the driver of the car has when it is being driven. As a driver, one would not expect the car to behave very differently just because the fuel is changed.

Following is a description of some of the things needed to be addressed when de- signing a CNG control system and it is based on the experience and knowledge within AVL.

4.4.1 Hardware related

CNG valves The CNG gas system in a car has at least one controllable valve between the high pressure tank and the injectors. The pressure in the tank can be as high as 200 bars, and when the engine is not running on CNG the valve is shut, to avoid having the high pressure gas in the fuel lines and thereby limiting the leakage if a line breaks. Therefore there need to be some functionality which controls this valve. The simplest would be to open it when CNG is requested and closing it as soon as there is a request to switch to gasoline fuel, or when the engine is shut down.

CNG sensors The ECU needs to read and evaluate the signals from the sensors placed throughout the engine and transfer them to the rest of the control system.

In order to calculate the CNG injection duration, both pressure and temperature in the CNG fuel rail are needed. Relative to the tank the pressure in the fuel rail is low at around 2 − 7 bars.

For evaluating leakage, and also the CNG tank fuel content/level, a high pressure (> 200 bars) sensor is needed in the high pressure part of the fuel system. To have a more exact value of the mass of fuel in the tank there could also be a temperature sensor.

CNG injectors There are two types of CNG injectors, high impedance and low

impedance. The two types of injectors have completely different physical properties

and therefore need different drivers in the ECU. The high impedance injectors are

(29)

for gasoline. The current through these injectors is not high, < 5 A and therefore there are no special requirements on the injector driver.

This simplicity also leads to a lower throughput, requiring higher rail pressure, in order to deliver the necessary fuel mass. Pressures around 7 bars are common with this type of injector. Due to the much simpler and cheaper injector driver the high impedance injector seems to be the choice of the OEMs and is commonly used in their CNG applications. [13]

The low impedance injector has a much higher current and need a peak and hold type of injector driver. That driver controls the current to the injector, aiming for a high start current to quickly open the injector and then reducing the current to hold it open.

Since the performance and accuracy of the low impedance injector is higher, it is the most commonly used injector in retrofit CNG systems even though it requires a more expensive driver. With the low impedance injector the CNG fuel rail pressure could be lower due to its higher performance, making it possible to empty the CNG fuel tank to a higher degree than with the high pressure needed with high impedance injectors.

The injection duration calculation is also affected by the type of injector. The com- pensation factors for opening and closing the injectors are different and can be han- dled by different calibrations. When calculating the mass flow through the injector there are two types of flow to consider: chocked and not chocked flow. When the rail pressure is high, more than about 2 times higher, compared to the inlet manifold pressure, then the flow will be chocked.[12] When the rail pressure is lower, the flow will not be chocked. The calculation of the flow is simpler and more straightforward when the flow is chocked. For the high impedance injector this means that the flow is always chocked and the calculation is thus simpler. On the other hand, for the low impedance injector it becomes more complicated due to the lower rail pressure used as the flow could be either chocked or not chocked depending on the pressure difference between fuel-rail and manifold.

CNG AFR and gas properties Different CNG and biogas types have different

stoichiometric AFR. The difference is much greater than the difference between gaso-

line qualities. Therefore there is a need of some kind of quality adaptation in the

system which detects a different gas quality and calculates the corrensponding stoi-

(30)

chiometric AFR or injection correction. Not only the AFR differs between different gas qualities but also the gas’ specific gas constant could be very different. Methane for an example has an AFR of 17.2 and a gas constant of 510, while a type of natural gas from the Netherlands has an AFR of 13,1 and a gas constant of 450 (SGC gas data 2012).

CNG emissions The emissions when running on CNG are similar, or lower than when running on gasoline with a fully warm engine. However, with the NEDC cycle, which is used to evaluate and test the emissions of cars in EU, the engine is started at around 25

^◦

C making cold start emissions about 90% of the total emissions during the test.

The most important factor of the cycle emissions is how long it takes for the three- way catalyst to reach a high enough temperature to reduce CO and HC and the accumulated amount of emissions before that. When running on gasoline about 50%

of all HC are oxidized in the catalyst at a temperature of around 350

^◦

C, but for CNG the temperature is significantly higher and 50% HC oxidation occurs around 450

^◦

C. The explanation for this is that CNG consists of mainly methane and thus the HC emissions will also be mainly methane. Methane is more stable hydrocarbon than most of the hydrocarbon found in gasoline exhausts and thus it requires a higher temperature to react with the oxygen. CO emissions require lower temperature to oxidize and there is no difference whether the fuel is CNG or gasoline.

Since a higher temperature is required to reduce HC emissions with CNG, the strat- egy to raise the catalyst temperature in the start has to be more aggressive.

With a gaseous fuel there will be no wall wetting in the inlet port, as with liq- uid fuel, and therefore the transient fuel compensation must be treated differently.

In theory there is no need for transient fuel corrections when running on gaseous fuel but since it is almost impossible to exactly measure the air mass going into the cylinder during a transient, a compensation factor, but different one from the one for gasoline, will be needed.

The gaseous fuel also requires different injection phasing compared to gasoline. The

strategy used at low and part loads for gasoline port fuel injection is an early injec-

tion. The gasoline fuel is injected on a closed inlet valve letting the fuel lay on the

hot valve, evaporating, before the valve is opened and the air and fuel is drawn into

the cylinder. With CNG the strategy has to be quite the opposite. CNG fuel should

(31)

be injected on an open inlet valve, waiting until the air passing the CNG injector has reached some speed, and thereby enhance the mixing between CNG and air. At low loads it is a common problem in CNG engines, using the same injection phasing as gasoline, that the mixing, between CNG and air, is so bad that misfires occur.

The misfires at low load could be harmful for the catalyst in the long run, as well as raising the HC emissions.

CNG cooling In gasoline engines, the cooling effect of the evaporating fuel is used actively to cool the combustion and lower the exhaust gas temperatures to levels that the exhaust valves, exhaust manifold, catalytic converter etc., can withstand. When temperatures come close to the design limit of a component, more fuel is injected so that the engine is running richer; as rich as Lambda = 0.8 can be reached at full load conditions. Gaseous fuels like CNG has no cooling effect. However since the fuel is persistent against knocking it is possible to run the combustion at an optimal phasing, even at high loads, thereby avoiding one of the reasons for high exhaust gas temperatures when running on gasoline.

Even though CNG has a high resistance to knocking, it might not be possible to run the engine at optimal ignition timing and combustion phasing. This since to the peak cylinder pressure reached during combustion could be higher than the design limit of the engine. If that should occur there are no other means of lowering the exhaust temperature than by reducing the load on the engine.

It should also be mentioned that if the engine is run with CNG, at same enrich- ment as gasoline, there is a very high risk of full load misfires. At a Lambda below 0.8 CNG very suddenly stops burning. These misfire will extremely rapidly destroy the catalyst where a big part of the unburnt CNG will be combusted reaching catalyst temperature in excess of 1500

^◦

C in matter of seconds.

CNG lubrication Gasoline normally has some lubrication effect on the valves and valve seals. A gaseous fuel, like CNG, being completely dry, has no such effect at all.

In combination with the lack of cooling effect and higher peak cylinder pressures, an accelerated wear of the valves and seals are common in CNG engines, especially those who are retrofitted from standard gasoline engines.

To prolong the life of the valves and seals a function in the CNG system can be

implemented, which at a certain interval forces some gasoline injections and thereby

lubricating the valves.

(32)

4.4.2 Drivability related

CNG filling Since CNG is a gaseous fuel, the fuel itself takes up more space in the cylinder than gasoline. This effect differs between CNG qualities but is around 10%. To have exact same drivability, as on gasoline, the engine management always need to compensate by opening the throttle more. This is normally handled by the torque calculations. But at full load there will be less torque with CNG.

CNG torque Torque will be different with CNG and is affected by a number of variables. The amount of air in the cylinder will be less with CNG causing the throt- tle position to be different for the same torque. The combustion property of CNG is different from gasoline. CNG is burns slower than gasoline but is also much more resistant to knocking, causing the ignition timing to be different. Enrichment, which also affect the torque, are not at all used with CNG as with gasoline.

To achieve a comparable level of drivability a complete separate torque calculation is needed for CNG. The model of how the torque is calculated could be the same, but all parameters and calibration data in the model need their own CNG specific values.

This is true whether the torque model is map based, or a parametrized model.

4.5 CNG Quality Adaptation

As discussed earlier there can be a quite large difference between CNG qualities. It is not only the AFR but also the gas constant of the fuel that can vary. Both the AFR and the gas constant will affect the fuel injection calculation. If the engine is not equipped with an AFR, the calculation of the air mass going into the engine is affected. There are different ways of handling this problem.

One way is to do nothing extra at all and to adapt the system by only letting the lambda controller adjust the difference and it will be saved as a fuel adaptation.

Even though this is an appealing way it has some drawbacks. The difference be-

tween fuels can be so large that it can be difficult to manage the adjustment within

the limits of the lambda controller. The idea of having the fuel adaptation in the

system is to correct for differences in the hardware of the components involved in

the fuel injection and lambda control, such as injectors and lambda sensors. If the

fuel adaptation shall handle both CNG quality differences and hardware, the risk of

going outside the limits and getting an engine malfunction light, will be much higher.

(33)

Another way is to do an extra CNG adaptation with the lambda controller off- set when refuel is detected and saving it as a CNG quality factor. In this way the problem with the limits of the lambda controller will be avoided.

Al of the above methods have the drawback that the effect of the gas quality on the air mass calculation will be completely ignored. The risk is also that even thouh the lambda controller will make sure that the lambda stays close to one att all times, an misscalculation of the airflow going into the engine will give an error in the cal- culated torque. This will in turn make the calculation for the ignition angle give a suboptimal ignition angle, and decrease the drivability of the car.

A third solution would be to have a more complex adaptive functionality which could

find both the stoichiometric AFR value and the value of the gas constant. If this

could be realized, then both the amount of fuel and the air would be calculated cor-

rectly and the complete system would work normally independent of the gas quality

used.

(34)

4.6 Rap2L overview

Rap2L is divided into tree main parts: Input, Engine Control Module (ECM) and Output. This division enables both an easy overview of the system and ensures that the same ECU can be used both in the development stage, on a MVEM, and later on a real engine.

Figure 5: Rap2L top page

4.6.1 Input

The Input function makes sure that all the signals end up on the right bus to the

ECM, regardless wether they come directly from hardware or are simulated signals

from the Output block.

(35)

4.6.2 ECM

In Figure 6 an overview of the top layer of the ECM is shown.

Sensor calculations As the amount and position of sensors that can be put on a real production car are highly limited, both by cost and available space, many of the sensor values have to be simulated. The sensor calculator function takes its signals directly from the Input block and from this tries to calculate a wide range of variables such as: the pressure before, and the mass flow over, the throttle, at what angle 10-, 50-, respective 90% of available fuel has been combusted and many others.

Limits When controlling an ICE it is very important to make sure that all of the parameters are within the tolerances set. As exceeding these limits can lead to engine failure, higher emissions and diminished driver experience, if allowed to continue.

Torque requirement and realization Requirement starts by taking the pedal angle, and vehicle speed, and translating it into a requested engine torque. It also makes sure that the engine speed does not go below the idle speed limit. This signal is then passed on to realization which translates it into the required air mass and spark efficiency.

Air Takes the calculated air mass and calculates the required cam angles and even- tual turbo boost pressure.

Fuel Calculates the amount of fuel to inject, how long time it takes to inject it as well as when to start injecting it. It also takes care of eventual enrichment due to cold start and lambda control.

Spark Can use both calibrated data or calculations in order to find the maximum brake torque ignition angle.

Aux, Cam, Boost and Throttle The ECM-module ends with hardware specific

functions that calculates the actual actuator values necessary.

(36)

4.6.3 Output

The Output has two different modes, either as a pure hardware interface, when running against a real engine, or as shown in Figure 7, where a full engine, with chassis and driver-model, are implemented.

MVEM In order to give a better overview of the system and make modifications easier, the MVEM in Rap2L has been constructed as a real engine where all the sub-modules corresponds to a physical component in the engine, and all the signals corresponds to mass or energy flows between them.

Airflow

Air filter, Inter cooler, Inlet manifold, Exhaust manifold and the Exhaust system are in this implementation modeled just as simple pressure differences as:

dP

_ds

dt = m · R · T ˙

P

_us

· V

_{AirF ilter}

(3)

where:

P

_us

= P

_Amb

+ 2.2 · 10

⁵

· ˙ m

²

(4) with P as pressure, ˙ m as mass flow, R as specific gas constant for air and V as ve- locity with subscript us, ds and Amb denoting upstream, downstream and ambient, respectively. The factor 2.2 · 10

⁵

is a calibration factor from empirical testing.

Throttle

The throttle is modeled as a venturi restriction of the flow by:

Q

_thr

= p

_us

· ψ · A

_{T hr}

√ R · T

_us

(5)

Where γ is used as the adiabatic index equal to

^C_C^p

v

:

ψ =



 

 

 

  s

2·γ γ−1

·

pds

pus

(

²γ

)

−

pds

pus

(

^γ+1γ

)

if

^p_p^ds

us

> p

_crit

=

2 γ+1

(

γ−1^γ

) s

2·γ γ−1

·

2 γ+1

(

γ−1²

)

−

2 γ+1

(

^γ+1γ−1

)

otherwise



 

 

 

(37)

Engine

The engine is modeled as a time delay that takes the engine speed and calculates how long time the air spends inside the engine before it is added with the fuel mass and expelled in the exhaust. The total mass flow in the exhaust is also used in order to calculate a very simple exhaust temperature model.

Combustion and emission Calculates the lambda value before and after the cat- alyst as well as the cylinder pressure during the full cycle.

Driver The driver model takes a reference velocity the car shuld follow and uses a combination of PID and feed-forward control, with a simplified model of the car, to generate the desired pedal positions and gear.

Manual transmission Takes the engine speed and desired gear from the driver function and calculates the rotational speed of the drive shaft.

Driveline model Uses a very simple model with rolling resistance, dynamic drag,

and engine/car inertia to calculate the vehicle velocity as a function of engine torque

and gear.

(38)

Figure 6: Overview of the original ECM top layer

(39)

Figure 7: Overview of the Output with plant model

(40)

5 Method of implementation

A deeper dive into Rap2L and the thoughts behind the changes done to it.

5.1 Design

In order to easily switch between using the system for running purely gasoline or enabling the use of alternative fuels, it was decided that a separate Alternative Fuel Adjustment (AFA) function should be created where all the calculations for CNG are made. In order to control whether the CNG part of the system is being run a system variable, e AltvFu, was created.

At each point where a signal needs to be altered a Goto-block is put which takes the original signal and feeds it to the AFA block. Another Goto is then used to feed the altered signal back and a choice is made, with the help of e AltvFu, whether to use the original or the adjusted signal. Shown in Fugure 8

(a) (b)

Figure 8: Layout of a choice

5.2 Variants of the ECM

At first the idea was to use buses as the main method to send the signals, as it is the

standard in other parts of Rap2L, as can be seen in Figure 9. This idea was later

changed to using Go-To block instead, making the system much more similar to the

original system, see Figure 10. This change both makes the system, as a whole,

cleaner and easier to follow, as well as increasing the portability of the system.

(41)

Figure 9: Overview of Rap2L ECM top layer, with buses.

Figure 10: Overview of Rap2L ECM top layer, with Go-To block implemented.

(42)

5.3 AFA

Figure 11 shows a overview of the top layer of the AFA function that is used for CNG calculations. An effort has been made in order to try to use of the same system layout as the main Rap2L.

Figure 11: Overview of Rap2L AFA top layer

5.3.1 Fuel Switch Manager

As different fuels give different torque for a given air mass and engine speed, the switch between fuels has to be gradual in order to not give a noticeable discomfort for the passengers in the vehicle. The two different ways of doing this are either to swap the fuels in one cylinder at a time, with time for the engine to adjust in between, or to gradually fade between the fuels on all cylinders simultaneously.

In order to successfully implement this second approach a fade factor, that con- trol the amount of air each fuel get assigned, was introduced. This also enables the lambda controller in the original Rap2L to remain the same, as it only tries to regulate the amount of unburnt oxygen in the exhaust.

One alternative approach that was considered was to give the full air mass to both

fuels and then use a global lambda to calculate the individual lambdas in order to

get the correct amount of fuel in the combustion, this idea was quickly abolished due

to the added complexity of calculations.

(43)

5.3.2 Fade

The function responsible for making sure that the desired and actual fuel are the same, checking that all requirements are fulfilled before a fuel switch request from the driver is implemented and whether a fuel switch is required for internal reasons.

If a fuel switch is required it also generates the Fade factor used to correctly switch between the fuels. This Fade factor is the variable defining what fraction of each of the fuels should be used, with 0 being pure gasoline and 1 being pure CNG.

5.3.3 Sensors

In order to calculate the volume the injected gas occupies in the manifold, and the subsequent changes in volumetric efficiency, a virtual sensory block was created.

From fuel, described further down, the the injector time is given. This enables the volume the CNG occupies in the manifold to be calculated by

V

_{CN G}

= t

_inj

V ˙

_inj

K

_voll

(6) where t is the time the injector is open, ˙ V

inj

is the volume flow of gas through the injector per unit time at the pressure and temperature in the fuel rail and

K

voll

= P

_Rail

P

_{M anif old}

T

_{M anif old}

T

_Rail

(7)

is the difference in volume that a gas occupies between rail and manifold, derived by P V = mRT → {m

₁

= m

₂

, R

₁

= R

₂

} →

^P_T¹^V¹

1

=

^P²_T^V²

2

→

^V_V¹

2

=

^T_T¹^P²

2P1

In order to keep all the calculations as simple and fast as possible the volumetric efficiency is calculated as the original volumetric efficiency from the original Rap2L, with the volume that the CNG occupies removed. This function is also responsible for calculating the maximum volumetric efficiency the engine would be able to acquire during the conditions that is at the moment.

5.3.4 Limits

As its twin for gasoline, it calculates the maximum and minimum values for a wide

range of variables depending on the operating point of the engine.

(44)

5.3.5 Request & Realization

Calculates idle speed, Torque-to-Fuel ratio and the efficiency of the combustion as a function of lambda. As calibration data for the Torque-to-Fuel ratio were not available for the engine at the time of completion, the same values as for gasoline were used but divided with 1.2. This give a acceptable value as long the ignition map is constant.[2] As the ignition is not held constant this gives a to low value, but acceptable as a first approximation.

5.3.6 Spark

Uses a look-up table to calculate the maximum brake torque angle.

5.3.7 Fuel

Responsible for calculating how much CNG is requested, calculates how much of the different fuels are actually injected with the help of the fade factor, the time it takes to inject the given amount of CNG and at what crank angle to start injecting CNG.

Fuel request Takes the requested gasoline fuel mass and calculates the requested CNG fuel mass and multiplies them with the control signal from the lambda PID- controller in order to get the actual requested fuel masses.

Fuel fraction Is responsible for taking the calculated fuel mass for both CNG and gasoline and weighting them with the factor given by the Fade function.

When weighting the fuel mass for gasoline it first removes any eventual enrichment added. If gasoline is supposed to be injected, this is then added back to the fuel mass.

Fuel cut If the driver suddenly releases the accelerator or the engine reaches it maximum rpm the fuel cut goes in and cuts the fuel supply in order to lower the fuel consumption when no power is needed, as well as protect the engine when maximum rpm is reached.

Mass to time Uses equation (8) to calculate the time it takes to inject the re-

quested fuel and at what phase angle the injection should start. The injection has

to be delayed sufficiently in order to avoid the turbulence just as the intake port is

opened, as well as have time to inject it all before the intake port is closed.

(45)

t

_inj

= P

_{Af tT hr}

K

_γ

√ T

_In

R

_{CN G}

1 M

_inj

(8)

where

K

γ

= v u u t

2γ γ − 1

2 γ + 1

2

γ−1

− 2

1 + γ

1+γ γ−1

!

5.4 Timing, computation and performance

One important concern when building this program was the fact that the time to

run a simulation should not increase significantly. At first the thought was to use

e AltvFu as an enable flag and thus disable the entire CNG-block when running a

strictly gasoline system, thus reducing the program back to its original size. It was

later decided that it is more efficient to put individual triggers on the subsystem with

matching period as the sister function in original Rap2L, and then use the enable to

decide if, and at what frequency, the triggered system should execute.

(46)

6 Results

The results are given.

6.1 Validation

When running the validation for this model the 200 first seconds of the New Euro- pean Driving Cycle (NEDC) were used. The required speed profile for this part of the NEDC can bee seen in Figure 12. [5] It is also important to point out that the look up tables used when running these simulations are not optimized for CNG and thus the numerical value differ from what will be when running on an actual engine, but trends can be seen. In all of the following plots green denote the CNG value while blue denotes the value corresponding to gasoline.

Figure 12: The profile for required speed.

(47)

6.1.1 Fuel

In Figure 13 the lambda value is shown for gasoline respective CNG, before the catalyst. As can be seen, the lambda controller manages to keep lambda around one for both fuels during steady state and around 10% error during transients. The extremely high lambdas that can be seen during deceleration is due to the fuel cut that happens during breaking, giving a infinite lambda. Since CNG uses the same limits as gasoline, the fuel cut that gasoline gets when the required air goes below the minimum air, does not happen for CNG. This can also be seen in Figure 14 as a plateau in the consumption for gasoline.

Figure 13: Lambda before the catalyst.

Legend: green - CNG, blue - gasoline, Red - speed requirement

(48)

Figure 14: Energy consumption during the drive cycle.

Legend: green - CNG, blue - gasoline, Red - speed requirement

white

(49)

6.1.2 Spark

In Figure 15 the spark angle when running on the different fuels is shown. As can be seen, the CNG can have a larger spark advance as it is not limited by knock.

The values calculated here is only based on the manifold pressure, and can thus be optimized further when access to an engine exists.

Figure 15: Spark advance for gasoline.

Legend: green - CNG, blue - gasoline, Red - speed requirement

(50)

6.1.3 Air

In Figure 16 and 17 the 7% increase in airflow needed for CNG, due to the decreased volumetric efficiency and increased AFR-ratio, is shown.

Figure 16: Airflow in intake manifold.

Legend: green - CNG, blue - gasoline, Red - speed requirement

(51)

6.1.4 Fade

In order to show the influence of switching fuel during driving a constant velocity of 50 km/h where chosen. The start up is done with CNG, and when a steady state has been reached, a fuel switch was ordered. In Figure 17 and Figure 18 it can be seen, between second 28-30, that the throttle closes and the airflow decreases.

Model Based Control Algorithm for a Bi-Fuel Engine with fuel adjustment