1.0 Introduction

As leaded petrol was phased out in various countries around the world, there arose a need for some technology that could perform as lead compounds once did in automotive engines. One such technology that could apparently do all that lead could and more was a stainless steel canister filled with small pebble sized lumps of tin. Rumoured to have been invented by a Royal Air Force technician in World War II, the technology has been developed by companies around the world and promoted as a panacea of internal combustion engine woes.

However, controversy has raged for many years over whether tin based fuel catalyst products actually work. They have been widely dismissed by the scientific establishment and yet the supporters of the technology can produce volumes of testimonial and anecdotal evidence provided by satisfied customers. The purpose of this research project is to add scientifically valid information into the public domain, which illustrates the effects if any, of installing a tin fuel catalyst device onto a spark ignition petrol engine. A secondary purpose is then to propose the mechanism by which any effects occur and finally, to suggest further work that will answer any remaining questions on the subject of tin based fuel catalyst products.

1.1 Background

Engine knock, pinging or detonation is a combustion phenomenon, which occurs in spark ignition engines operating at high temperature or high load conditions. It is considered to be an undesirable effect as serious damage to the piston crown and other internal parts can be the result. The risk of this damage means that knock limits the operating parameters, the design and ultimately the performance of an engine. One of the parameters that most affects the onset of knock is the “octane rating” of the fuel. This is a measure, relative to iso-octane, of the tendency for the fuel to induce knock. A higher octane fuel will allow a given engine to operate with more aggressive compression ratio and ignition timing without knocking, thus producing a higher power output. Since 1922 tetra ethyl lead has been added to petrol to improve octane rating and engine performance. Charles F. Kettering and Thomas Midgely, the men who originally discovered the benefits of tetra ethyl lead (TEL), also investigated iodine, iron, nickel and tin compounds but found that none were as effective as lead at reducing knock in internal combustion engines [1]. Tetra ethyl tin (TET) was found to be only 13% as effective as TEL as a knock suppressant [2]. In addition to increasing octane, TEL will also leave deposits on the sealing surfaces of valves and valve seats, which will protect them from excessive wear [3]. Unfortunately the unacceptable health risks associated with the use of lead prohibits its continued use. With the worldwide trend away from leaded petrol since the late 1980s, the need for valve seat recession (VSR) protection became paramount, especially for older vehicles designed to run on leaded petrol. High octane unleaded petrol uses aromatic compounds such as benzene in place of lead organics, but does not offer any VSR protection. The introduction of unleaded high-octane petrol meant that owners of older vehicles would have to use a lubricating additive to protect the valve seats of their engines. Tin fuel catalyst devices were one range of products marketed as a solution to this problem.

1.1.1 History

The origins of tin fuel catalysts are rumoured to be in World War II Russia. Legend has it that a Royal Air Force aircraft technician working on fighters stationed in Murmansk discovered the solution to the problems associated with the low grade Russian aviation fuel. Sergeant Henry Broquet’s tin catalyst technology reportedly allowed the British aircraft to perform as well as they had done on leaded RAF fuel. Broquet emigrated to South Africa after the war, taking the secrets of his technology with him. By 1989 a company bearing his name was marketing the first of the modern variants in South Africa and the United Kingdom [4]. The phasing out of leaded petrol in many countries at the time established a market for products that could prevent VSR, not to mention the many other benefits attributed to tin catalyst products. Consequently a number of small companies took up the technology for themselves. The more prominent tin catalyst brands include: Broquet, Carbonflo, Greenpower, Doring, Fitch, Comtec, Fuelcat, Ultraburn, Fuelsaver, Powerplus and in New Zealand; Fuelstar.

The products usually take the form of a stainless steel canister as pictured in Figure 1.1, which is connected in-line with the fuel supply to the engine. Inside the canister, the fuel flows over pellets or cones of various tin alloys. The alloy is often described as an alloy of tin and antimony [5]. In some cases magnets are also employed, with the suggestion that the paramagnetic qualities of the fuel are utilised to benefit engine performance [6]. Usually the tin pellets are separated by perforated steel plates, which presumably tailor the fuel path around the pellets as well as providing a harder material for the tin pellets to wear against. In addition to the canister type, some manufacturers offer a steel mesh bag or a short steel tube containing tin pellets which is then dropped into the vehicle's fuel tank.

Figure 1.1. Fuelstar PS110 canister

In addition to the VSR protection, tin catalyst manufacturers often make claims of 10% fuel savings, 10% power increase, 50% reduced toxic emissions and overall improvements in engine running characteristics such as smoothness, throttle response, starting, and lifespan [7].

The suggestion that they can overcome the failings of older vehicle engines, which were designed to run on leaded fuel, has been controversial as these claims have mostly gone unsubstantiated. In fact, several manufacturers have been penalised for producing advertisements, which make claims based on hundreds of testimonials from satisfied customers, yet cannot be backed up with scientifically valid testing [8]. There is currently little conclusive information available regarding the performance of these products as much of the testing that has been performed in the past is either privately owned information or has disputed validity. The International Tin Research Institute and www.tintechnology.com has undertaken its own research, but charges a large fee for access to their work [4]. However Dr. Jeremy Pearce has generously provided some very useful information from that body of work. Unfortunately many of the information sources quoted by manufacturers from around the world have proven to be only vaguely relevant and some appear to be entirely fictional. New Zealand manufacturer, Fuelstar International Ltd has also commissioned research on their technology, some at the University of Auckland and this work has proved invaluable in preparing this report.

1.1.2 Tin properties.

Tin is a soft shiny metal, often used to plate steel and is the major component of electrical solder. Its solubility in hydrocarbons such as petrol is negligible, but it will form a particulate suspension of charged clusters of metallic ions known as colloids. A colloidal suspension contains particles that measure around 100nm or less. Several researchers have found that a vibrating tin catalyst canister is capable of producing such a suspension [9,10]. Particles in the nanometre size range exhibit unusual behaviour as a result of the enormously increased surface area with respect to mass, including magnified catalytic effects. Colloidal suspensions of tin and antimony and their oxides are regularly employed in industry as catalysts and flame retardant additives in polymers. Tin oxides are used as catalysts in coal liquefaction in gas sensors. The main automotive application of tin is as a component of “white metal” and “Babbitt” bearings. Tin-lead-antimony alloys are effective solid lubricants and are often used with engine crankshafts and conrods. Alloys of this nature are soft enough that small particles will not damage the internal parts of an engine should they enter the combustion chamber or crankcase.

The tin alloy commonly used in tin fuel catalyst products is composed of approximately 70% tin, 10% antimony with the remainder including unspecified quantities of lead and mercury [9]. Tin in its metallic form is generally considered to be non-toxic. This is clearly illustrated by the use of tin plated steel in food packaging. Tin is becoming popular as a lead replacement metal in a variety of industries as a result of its relatively safe nature. Compounds of tin including its oxides however, are significantly less benign. Sustained exposure to tin oxides can result in skin and membrane irritation and if inhaled in sufficient quantities, pneumoconiosis, a mild lung inflammation may occur [11]. Tributyltin (TBT) is a highly toxic tin compound, which has been used for many years as an antifouling product on marine vessels. The detrimental environmental effect of this product is such that the International Maritime Organization is moving to totally ban tin based antifouling products on all marine vessels [12]. Antimony salts are known to have effects similar to arsenic poisoning, while antimony trichloride is considered corrosive to human tissue [11]. The health hazards associated with lead and mercury are widely known and both metals are radically more harmful than either tin or antimony.

Other metals are known to have a detrimental catalytic effect on petrol, notably copper and nickel and a non-metal, sulphur, which are all found in fuel and fuel systems. Exposure to these elements can promote oxidation and sulphation of the fuel, which generally diminishes the effectiveness of the fuel [13].

1.1.3 Theoretical Mechanisms

There are two main theories on how the tin affects engine performance. The most commonly suggested theory is that microscopic colloidal particles of metallic tin are deposited into the fuel stream and then travel with the fuel into the combustion chamber where they can affect the flame and also leave protective deposits on the piston and valves [9,14]. The prerequisite of this mechanism is vibration of the canister, usually attained by mounting it on the engine itself. As the loose tin alloy cones, visible in the drawing Figure 1.2, bounce and slide across the steel plates between them, a minute amount of the alloy is worn off and carried away by the fuel stream flowing through the canister. It has been observed [9] that some of this suspended material can precipitate onto the surface of fuel lines and filters, thus preventing its entry into the combustion chamber of the engine. In order to prevent this phenomenon affecting their product’s performance, Fuelstar specifies in their installation instructions [Appendix J] that downstream hoses and pipes are to be cleaned with a suitable solvent and that all fuel filters downstream of the canister be removed or relocated.

Figure 1.2. Typical interior of a tin catalyst canister

The action of colloidal particles once they arrive inside the combustion chamber is uncertain. It has been suggested that metallic and oxide particles play a part in modifying the combustion processes in the same way that lead compounds do in leaded petrol. The catalytic effects of tin and antimony are fairly well established and industrial applications such as flame retardation and coal liquefaction. Metallic tin particles may catalyse the combustion reactions and somehow slow the rate of combustion. This would yield an improvement in apparent octane rating and thus knock reduction. Existing tin oxide particles along with those created from oxidation of metallic particles may then alter the combustion further, resulting in improved efficiency and reduced toxic emissions [9,14]. It is also proposed that a residual accumulation of tin and tin oxide on the internal surfaces of the engine contribute to a progressive increase in these effects. Testimonial reports from satisfied users regularly mention a grey or white deposit on piston crowns and occasionally on exhaust pipes, which is often touted as evidence of these reactions.

The second viable hypothesis suggests that the transport of tin particles into the combustion chamber is not the fundamental mechanism. Rather, that the surface of the tin alloy pellets inside the canister catalyse the precipitation of metal oxides and sulphides from the fuel, thus improving the fuel’s combustion properties. Sulphation and oxidation of petrol is known to be detrimental to fuel quality [13,15], so the reduction of these species must in this case be beneficial to the engine’s performance. Most tin catalyst manufacturers suggest that canister vibration is a requirement for effective operation of their products. The effect of this vibration and the friction and abrasion between internal parts may be that a clean surface of tin alloy is regularly exposed to the fuel flow. Should this be the case, any tin released into the fuel stream is likely to be waste effluent in the form of oxides and sulphides. [15]. There remains the possibility of course, that this effluent also contributes some beneficial effect to engine performance besides merely denying the opportunity for oxidisation and sulphation of the fuel.

If in fact, these products do not work at all, the benefits described in testimonials from satisfied customers are possibly perceived benefits as a result of some kind of “placebo effect” [16]. The placebo effect is often used in medical research to determine if the effectiveness of a new treatment is purely a result of the treatment alone, or whether the mindset of the patient is contributing to their improved wellbeing. The aftermarket nature of these products often means that they are purchased by people who have already identified a need to reduce their vehicle’s fuel consumption or improve its performance. It has been suggested that the driver’s awareness of this need can lead to a slightly less aggressive driving style whether consciously or not. This will also have the effect of improving fuel economy and reducing the amount of time the engine is operated in knock susceptible conditions. The existence of such an effect could explain the abundance of excellent testimonial evidence for tin catalysts, which cannot be replicated in laboratory conditions.

In medical research the placebo effect is evaluated by applying a blind or double blind test. In a blind test some proportion of the sample population is administered a known ineffective treatment while the remainder get the experimental treatment. In each case the patients are not aware of which they have been given. In order to further guarantee the integrity of the experiment, a double blind test may be performed whereby neither the patient or the person administering the treatment is aware of which they are using.

1.2 Project approach

The objective of this project has been to evaluate the effectiveness of tin alloy fuel catalyst products and if they can be shown to work, to propose the mechanism by which they operate.

The first task was to conduct a literature review covering the nature of the products, their claimed benefits and some scientific basis for their usefulness. The mechanism by which leaded fuel and other additives affect engine performance has also been considered. It is likely that if tin alloys do have an effect, that they work in a similar way to lead compounds. Based on this information and the various hypotheses of how tin affects performance, a schedule of experiments was devised to examine some of the claimed effects. The experiments themselves were designed to give an indication of which mechanisms are most responsible for the attractive improvements in vehicle performance experienced by many satisfied users.

The schedule of experiments chosen covers three areas. Laboratory engine testing of output and knock behaviour, chemical analysis of fuel and combustion deposits and road trials of vehicles fitted with tin catalyst and placebo canisters. Each experimental method has been carefully considered so that technical deficiencies claimed in previous work are not repeated here.

The results and conclusions reached through this project should clarify some of the facts relating to tin based fuel catalysts and allow a fair assessment of the real benefits associated with the products.

2.0 Literature Review

A literature review was conducted in the initial stages of this project, which focussed on researching the products, their manufacturers and their claims and on establishing the generally accepted mechanisms by which traditional fuel additives work. A great deal more effort was required to find useful information relating directly to the scientific principles that apply to tin catalyst products. While some scientifically valid investigations have been undertaken, many of those are unattainable privately owned information, while others are of disputed accuracy. Most of the manufacturers can provide only testimonial evidence or data that cannot be validated.

2.1 Earlier Testing of Tin Catalyst Products

The literature survey has revealed several sources of research into the subject in the past. Locally, Fuelstar has commissioned reports by the University of Auckland’s Energy and Fuel Research Unit, Dr. T James Sprott [17] and chartered chemist Ron Wilkinson formerly of CSIRO and Australian Fuel Developments in Melbourne, Australia.

· R. A. WILKINSON: Mr. Wilkinson performed tests on the Fuelstar product to measure any changes in the fuel chemistry after having been passed through the device. He found that there were no detectable effects with a single pass or without vibration, however when the fuel was recirculated for 20 minutes and more, small amounts of tin were detected in the fuel. The metallic tin was in the form of microscopic colloidal particles suspended in the liquid fuel [9].

· EFRU: Fuelstar NZ commissioned tests on their product by the Energy and Fuels Research Unit of the University of Auckland. The results indicate some small changes, but no significant improvements as a result of the fitment of the Fuelstar device [18].

· NZAA: The NZAA commissioned Prof. Harry Watson of the University of Melbourne to conduct tests on a Fuelstar device to AS4430.1 1996. These tests concluded that there was no significant effect on knock or VSR in the test engine [19]. These results have since been disputed by Fuelstar NZ and Dr. T James Sprott on the basis that the standard was not followed closely [17].

· UTAC: This French organisation found that relatively large quantities of tin alloy could be delivered into the fuel stream with adequate vibration. Fuel flow rate appeared not to be a significant factor, but filtration did remove a large proportion of the added metal [10].

· RAC The Royal Auto Club UK was commissioned by the Advertising Standards Authority and the Office of Fair Trading in the UK to report on the claims made regarding the Fuelcat product. The RAC concluded that Fuelcat could not substantiate their claims of improved engine performance and consequently Fuelcat was censured by those government agencies [8].

· FBHVC: The Federation of British Historic Vehicle Clubs performed tests on a number of products marketed as solutions to problems expected due to the introduction of lead free petrol in the UK. Of the 12 products submitted, only four provided suitable improvements in valve seat recession. No tin canister products were successful [20].

· ITRI: The International Tin Research Institute of the UK has engaged in research on the subject of tin and its effects on engine and fuel performance. ITRI charges approximately NZ$8,000 for full access to this information [4]. Advice from Dr. Jeremy Pearce of ITRI has proved very useful in designing the experimental procedures.

2.2 Research into Mechanisms and Theories

In order to devise a suitable set of experiments to evaluate the effectiveness of these devices, it is helpful to understand how the traditional additives achieve the same effects. Tetra ethyl lead is the most important of the fuel additives used in the past and a great deal of research has been undertaken into its effects, mechanisms and the effects of eliminating it.

Traditionally, tetra ethyl lead (TEL) has been used for protection against valve seat recession (VSR) and knock. Along with its discovery in 1921, several other compounds were shown to benefit engine performance, particularly; iron carbonyl, iodine and tetra ethyl tin (TET) [1].

Since the phasing out of leaded fuel, the most used petrol additive packages contain compounds of potassium, phosphorus and sodium for protection against VSR. Manganese containing Methylcyclopentadienyl Manganese Tricarbonyl (MMT) and the fully organic compound Methyl Ter-Butyl Ether (MTBE) are also used in some countries to improve octane ratings and reduce exhaust emissions, but several authorities have restricted or banned its use [21]. A higher proportion of aromatic a hydrocarbons is now the typical method for achieving high octane ratings.

2.3 Knock

Knock, pinging or detonation is a combustion phenomenon, which occurs in spark ignition engines operating at high temperature or high load conditions. It is generally considered to be an undesirable effect as serious damage to the piston and other internal parts can be the result.

The phenomenon occurs when the normal fuel air combustion deteriorates from a stable flame to a small and uncontrolled explosion. These occur mostly near the cylinder walls and the edges of the piston crown, often blasting off or melting small pieces of material. As shock waves from the explosions bounce from one side of the combustion chamber to the other, a pinging or knocking sound is produced, hence the name.

A lean mixture, excessive ignition advance, excessive compression, excessive combustion chamber deposits, insufficient cooling and insufficient fuel octane rating can all cause knock. The differences in pressure between normal combustion, pre-ignition and knock are depicted in Figure 2.1. The ragged edge of the knock waveform represents the high frequency explosion shockwaves reflecting off the cylinder walls.

Figure 2.1. Pressure traces of normal, pre-ignition and knock

There are two theories that describe exactly how knock may occur: [5]

· Autoignition: Normal ignition results in a flame front propagating outwards from the spark plug. As the flame front advances towards the walls of the combustion chamber the pressure & temperature of the un-burnt “end gas” in front of the flame wall increases. If the pressure & temperature of the end gas is sufficiently increased then a secondary ignition within the end gas itself can result, causing knock.

· Detonation: If conditions are such that the flame front reaches the speed of sound as it propagates into the end gas then the rate of energy released is so great it causes these localised pressure increases in the end gas. Detonation occurs at these points.

2.4 Effects of Knock

Knock effectively presents a limit to the operating parameters of an engine. Fuel/air ratio, ignition timing, cooling, and octane requirement for a given engine must all be selected to avoid the occurrence of knock, as engine failure would otherwise be the result. The small explosions will cause substantial erosion inside the combustion chamber if allowed to continue and the high local temperatures can melt the aluminium components such as piston crowns or cause pitting in valve faces and spark plugs. Figure 2.2 illustrated the possible extent of knock damage. Piston rings gaskets and bearings can also be badly damaged as a result of the shockwaves induced [11].

Text Box: Figure 2.2. Knock damaged piston

2.5 Valve Seat Recession (VSR)

One of the more important claims made by manufacturers of tin additives is the reduction of valve seat recession (VSR). The extreme temperatures experienced, particularly by the exhaust valve seating surfaces cause microscopic welding or adhesion between the valve and its seat each time the valve is closed. Upon re-opening of the valve, tiny particles, which have become welded to one surface, are torn from the other. In addition to this action, the relative rotation of valve and seat, coupled with the abrasive adherent particles causes further erosion of the mating surfaces. As a result of this material removal, the valve gradually recedes into the head [3]. The damage to the sealing surfaces can promote burnt valves, local overheating and necessitate early valve clearance adjustments. Eventually the damage will render the cylinder head irreparable.

Tetra ethyl lead (TEL) has been the most used additive to prevent this phenomenon in engines. It is believed to produce particles of metallic lead and lead oxide (PbO), which then precipitate onto the hot valve surfaces. The oxide “glaze” and the lubricating effects of the soft lead then prevent contact between the weldable steel surfaces, hence preventing further erosion [3].

Tetra ethyl tin, phosphorus, sodium and potassium based products are believed to work in the same way [22].

Modern engines utilise hardened steel or ceramic valve seat inserts to prevent excessive VSR, but older vehicles with soft cast iron seats are still subject to damage without added protection.

2.6 Leaded Petrol

General Motors employee, Charles F. Kettering discovered in 1921 that small amounts of tetra ethyl lead (TEL) could significantly increase the octane rating of petrol [23]. During the course of his experiments tetra ethyl tin (TET) and other compounds were tested, but found to be much less effective [5]. An unexpected side effect to the addition of TEL was the protection from VSR, which allowed the use of the unhardened cast iron of the head to be machined for the valve seats. TEL was used exclusively worldwide in petrol until concern over the health hazards associated with lead became sufficient to force a change to unleaded fuels for all but special purpose applications.

2.7 TEL Knock Mechanism

The mechanism by which TEL prevents knock is not exactly known, but there are several theories that can explain it. It is suggested that the presence of TEL in the combustion chamber will slow the rate of combustion and flame velocity and/or raise the critical temperature and pressure at which knock will occur. This may be a result of the TEL particles absorbing or delaying a portion of the combustion energy to break themselves down to hydrocarbons and lead compounds.

TEL is the metal-hydrocarbon compound with the highest molecular mass that is also petrol soluble. This may be the reason that TEL is so much more effective than other lower mass compounds. [8]

TEL also causes a number of undesirable deposits inside an engine. Leaded deposits can continue to grow to the point where piston rings stick and spark plugs become fouled. Exhaust valves which benefit from low levels of lead deposition will quickly burn if thicker deposits begin to flake off allowing hot gases to vent through the cracks [24]. Flakes from lead deposits also cause localised hotspots, which accelerate pre-ignition, and detonation. Scavenger compounds such as ethylene dibromide must be added to the fuel to help dissipate the excess lead deposits. Lead is then exhausted into the environment mostly as lead dibromide [8,24].

Lead compounds also cause significant degradation of engine oil and can accelerate corrosion of critical parts, especially hot exhaust valves [24]. Catalytic converters and oxygen sensors rely on precious metals to function, but the lead in exhaust gases will quickly poison the precious metal components and cause premature failure [25].

3.0 Methods

Three sets of experimental work were performed. Firstly, a Ricardo research engine was used to compare the performance of various treated fuels. Secondly, the different fuels and their combustion deposits were examined for the presence of tin. Thirdly, five vehicles took part in a road trial to compare their performance with and without a canister fitted.

3.1 Fuel Preparation

Text Box: Figure 3.1. Fuelstar (left) fitted to the Ricardo engine

There were four different fuel types used during the laboratory experiments. The base fuel was BP 91 octane petrol purchased in February 2002 and topped up on 6 September 2002. This petrol was stored and delivered to the engine through the Thermodynamics Laboratory fuel system. Tests of the Fuelstar PS110 canister used the same system, with the canister attached to the vibrating exhaust system frame shown in Figure 3.1. The base fuel was also used for the remaining experiments. Petrol for further treatment was taken from the delivery line immediately before the carburettor, to ensure that it had been exposed to the same environment as the petrol used in baseline and Fuelstar experiments.

Text Box: Figure 3.2. Tin cones and granules

The first modified fuel was treated with granular tin and filtered before use. Two tin alloy cones from a Fuelstar TM-VC unit were granulated with a coarse steel rasp (Figure 3.2). The resulting chips weighed 16.7g in total with an estimated surface area of 0.0318m². This is nearly three times the surface area of tin alloy in a PS110 canister. The treatment vessel was a sealed four-litre tin-plated steel petrol can. This vessel was chosen to minimise exposure to materials other than those found inside a Fuelstar canister. The petrol and granules were regularly agitated and stirred over a period of 20 hours. The fuel was then filtered to remove as much metal as possible. The fuel was passed twice through pleated Whatman grade 4 and grade 5 filter paper, ensuring 100% retention of particles over 2.5um and some retention of particles below that size. The petrol flowed directly from the can into the filter paper and into a brown glass flask. The funnel was covered with plastic film to trap vapours and reduce additional air exposure. A sample of both the filtered and unfiltered petrol was taken for analysis.

This fuel type is referred to as “granular filtered”.

Text Box: Figure 3.3. Abrasive, tin cone and suspension

“Abrasion suspension” fuel was a mixture of base petrol and ground particles of tin alloy. A quarter sheet of 1500 grit silicon carbide paper was placed in a large glass Petrie dish with a small amount of petrol. The Fuelstar cone was rubbed over the abrasive paper for a period of about 20 minutes, producing a dense grey suspension of metal in the petrol (Figure 3.3). The masses of the cone and the apparatus were monitored to determine the mass transfer. Approximately 0.3g of tin alloy was transferred into suspension with 100mL of petrol. This was diluted to 3 litres with clean base petrol giving a concentration of approximately 0.1g/L. A sample of this fuel was taken for analysis.

Figure 3.4. Friction apparatus. Drill press (right), new tip (centre)and worn tip (left)

“Friction suspension” fuel was produced by mechanically rubbing a Fuelstar cone against a steel plate while submerged in petrol. The slotted tin alloy cone was attached to a stainless steel shaft, (Figure 3.4) which was rotated at 280rpm by a drill press machine. With a maximum mass of 4.2 kg on the press handle, the cone was pressed onto the plate with a force of 410N.

The steel plate was soldered to the bottom of a four-litre tin-plated steel can with a small hole in its lid for the rotating shaft. After three hours treatment, only the top three litres of treated petrol were tapped to prevent large metal particles at the bottom of the can from being ingested into the engine. A sample was taken in a beaker to observe any particle precipitation and a second sample was taken for tin concentration analysis.

3.2 Ricardo engine experiments

The laboratory experiment series consisted of a set of Knock Limited Spark Advance (KLSA) tests, comparing the performance of a Ricardo research engine when run on standard BP 91 octane petrol, with its performance on petrol modified with tin alloy. The Ricardo research engine, produced by Ricardo and Company, Engineers Ltd. of England, is a 506cc single cylinder four-stroke stationary internal combustion engine with two overhead valves. Figure 3.5 shows both sides of the Ricardo research engine.

Figure 3.5. Ricardo engine, left and right sides

It can be operated on diesel, alcohol, CNG or LPG and for these experiments it was run on petrol. Compression ratio is adjustable from 7:1 to 22:1; ignition timing is adjustable from zero to 60 degrees advance before top dead centre (TDC) and air/fuel mixture can be set by a calibrated needle valve on the carburettor. With care each of these parameters can be adjusted while the engine is running. The internal components of a Ricardo engine are extremely sturdy, allowing it to be operated under conditions that would severely damage an ordinary automotive engine. In particular, the adjustable ignition and compression means that the engine can be used to examine knock behaviour for extended periods without suffering damage.

The air intake system fitted to the Ricardo engine incorporates air flow meters and a filter that draws air from within the laboratory. The exhaust system is supported from the floor and the ceiling and feeds into the ventilated exhaust duct overhead. Water-cooling of the cylinder head is controlled by a water/water heat exchanger. The Ricardo engine is mounted on a concrete pedestal along with its dynamometer in the University of Auckland, School of Engineering Thermodynamics Laboratory.

The engine is fitted with a Bosch HEI transistorised ignition system, firing one Bosch W3AC spark plug. The calibration of the ignition adjustment scale was checked by comparing the zero degree mark and the flywheel TDC mark using an adjustable stroboscope. The calibration at 20 degrees and 60 degrees advance was also confirmed.

Text Box: Figure 3.6. Ricardo engine and dynamometer

A Laurence, Scott and Electromotors LA16 eddy current dynamometer is directly coupled to the output shaft of the Ricardo engine as illustrated in Figure 3.6. Electrical power is applied to the dynamometer to start the engine, but once running the generated electrical output of the dynamometer is monitored and controlled by the PC based control system to govern the running speed of the engine.

A water-cooled Kistler 601A pressure transducer linked to a Kistler 5007 charge amplifier detects combustion chamber pressure. Engine speed and crank angle are measured with a Hall effect sensor reading from the engine flywheel and a K₃-type thermocouple senses exhaust gas temperature in the exhaust manifold at a point approximately 200mm from the valve.

Text Box: Figure 3.7. Ricardo engine control and data acquisition computers

Two PC compatible computers shown in Figure 3.7 monitored these parameters along with water temperature, oil temperature and the dynamometer output. The first computer was dedicated to control of the Ricardo engine via a Siemens SIMOREG control unit and to collection of basic data in Microsoft Excel. The second computer, running LabVIEW v5.1 and a Metrabyte DASK-16F high-speed analogue to digital converter card collected the pressure data at high speed, allowing further analysis of combustion and knock. During operation of the engine the pressure signal was monitored on a Nicolet 3901 oscilloscope to identify the intensity of knock in real time and avoid damaging the engine.

The Thermodynamics Laboratory features an integral fuel delivery system that feeds the laboratory engines. BP 91 octane unleaded petrol is stored in an underground tank, and pumped, as required to an external header tank and on to each engine’s fuel control system and filter. For all warm-up and baseline runs on 91-octane petrol this system was utilised. For tests of the standard Fuelstar canister, this system was diverted immediately before the carburettor, running the fuel hose to the canister and then back to the carburettor. As specified in Fuelstar’s installation instructions [Appendix J], the PS110 canister was mounted vertically to a vibrating structure. During operation of the engine it was noted that the most vigorous vibration occurred in the exhaust system support stand beside the engine. The horizontal vibration had a measured amplitude of approximately 0.8mm in total with the frequency corresponding to the engine operating speed. In the case of test runs at 1500rpm and 2000rpm this is a frequency of 25Hz and 33Hz respectively. Maximum oscillation occurred at 1700rpm (28Hz) with total amplitude of approximately 2.0mm.

For the modified fuels, an auxiliary fuel delivery system was devised with a four-litre portable steel tank was attached to the extended exhaust system stand. Vibration of the exhaust stand was sufficient to maintain the particle suspension of the fuels (Figure 3.8).

Figure 3.8. Ricardo and Fuelstar PS110 (left), and white auxiliary fuel tank (right)

A one-metre length of Goodyear rubber fuel hose connected the tank directly to the carburettor below. In accordance with Fuelstar advice, no filters were fitted to this system that could inhibit the transport of tin particles to the engine. After each run the tank was removed and cleaned.

To establish an appropriate air/fuel ratio for the experiments an Autodiagnostics ADS 9000-P five gas analyser was used. The exhaust gas was continuously sampled from a fitting in the exhaust system about one metre from the engine. The engine was operated at each of the chosen test speeds while the needle jet was progressively closed. Air/Fuel ratios and carbon monoxide (CO) levels were recorded at each needle jet setting to determine an appropriate needle position for the entire testing schedule. The Ricardo engine was set up to approximate the operating parameters of an average road driven car [25]. Compression ratio, running speeds, fuel mixture, water and oil temperatures are given in Table 1.

Table1. Ricardo engine operating parameters

Compression Ratio	8.5:1
Parameter	Setting
Throttle Position	100%
Running Speeds	1500 and 2000rpm
Needle Position	0.5
Air/Fuel Ratio	12.28 @ 1500rpm 13.55 @ 2000rpm
Water Temperature	60 ± 5 ºC
Oil Temperature	60 ± 5 ºC

If the Ricardo engine is considered as one single cylinder of a 2 litre, 4-cylinder engine or from a 3 litre, 6-cylinder engine, the running speeds of 1500 and 2000 rpm correspond approximately to driving speeds of 60 and 80 km/h in 4^th gear. The needle position of 0.5 was selected to produce CO emissions of around 3-5%, similar to those in road vehicles [25]. The actual measured CO level was 6.5% at 1500 rpm and 12 degrees ignition advance; and 3.2% at 2000 rpm and 12 degrees ignition advance. At 100%, wide-open throttle (WOT) knock is most readily induced, so this was chosen as the position where knock was most likely to cause problems. Due to the highly effective cooling system fitted to the Ricardo engine, temperatures of over 80ºC were impossible to maintain. It was found that the engine would maintain a temperature of around 60ºC without much adjustment. Temperatures of this magnitude are common in two stroke engines petrol engines, while four stroke engines usually run at around 80ºC.

3.2.1 Ricardo Procedure

Knock limited spark advance (KLSA) tests are a means of measuring the torque output of an engine up to the point where knock limits further advancement of the ignition timing. For a given compression ratio, air/fuel ratio and throttle position; the torque curves produced by running the engine on any fuel can be compared to those produced by running on the standard control fuel. In addition, as ignition advances, the onset of knock can be examined to evaluate any changes. A KLSA test was performed for each of the modified fuels immediately preceded or followed by a baseline test on untreated 91-octane petrol to provide comparative data.

The basic operation procedure for the KLSA tests is given below:

Oil warm-up. The engine oil heater is switched on until the oil temperature exceeds 30ºC.
Setting checks. Engine operating parameters are checked and re-set and the warm up spark plug is fitted. Set 30% throttle and 0º ignition advance.
Engine Starting. The engine control program is set to 600rpm and the dynamometer accelerates the Ricardo engine. Once up to speed the fuel and ignition are turned on and the engine runs under its own power.
Engine warm-up. Engine speed is set to 2000rpm. As the engine warms up, the throttle is opened to 100%. The engine is ready when the water and oil temperatures reach 60ºC. Atmospheric conditions are recorded at the air intake.
2000rpm baseline KLSA. Starting from 0º, the torque and exhaust gas temperature (EGT) are recorded at 0, 5, 10, 15, 17, 18, 19, 20, 21, 22 and 23 degrees ignition advance. The engine is run for 2 minutes at each setting to ensure stability before recording the data. Pressure trace data is recorded from the 17º point onwards.
1500rpm baseline KLSA. At the conclusion of the 2000rpm test the speed setting is reduced to 1500rpm and the engine is run for at least 7 minutes to allow the speed to stabilise. Starting from 0º the torque and EGT is recorded at 0,5,10,12,13,14,15,16 and 17 degrees ignition advance. Pressure data is recorded from 12º onwards.
Stop engine. At the conclusion of the 1500rpm run the speed is set to zero and the ignition and fuel supply are turned off as the engine decelerates. Once stopped, the spark plug and fuel hose are removed.
Inspect combustion deposits. When necessary the inside of the combustion chamber can be inspected through the spark plug hole with a boroscope and the appearance recorded. The condition and colour of the spark plug is also recorded.
Change fuel. A new sparkplug is fitted to the Ricardo engine. The auxiliary fuel tank hose is now connected to the carburettor to deliver the petrol to be tested.
The KLSA process is repeated from step 2 for the test petrol.

At the conclusion of KLSA runs with modified fuel, the engine was run at 2000 rpm and 12º advance until all the fuel was consumed to build a suitable combustion deposit on the sparkplug for analysis of metal content On occasions where there were atmospheric or running differences between the baseline and test fuel runs, a second baseline run on 91 octane petrol was performed. Manual recording of basic data was also performed as a backup and for immediate reference. Audible knock intensity and the knock patterns observed on the oscilloscope were assessed as the experiments proceeded. When knock was occasionally audible and visible on the oscilloscope, “MK” for minor knock was noted beside the record for that point. As spark advance increased to the point when observed knock intensity was obviously excessive, “KL” for knock limit was noted.

3.3 Ricardo Data Analysis

From the spark advance, torque and EGT information collected from the Ricardo engine in Excel, a spreadsheet was produced to analyst, tabulate and graph the data. Plots of torque versus spark advance for each fuel and its respective baseline run illustrate the difference in output as a result of the differences in the fuel. The last point of data may also be considered as beyond the knock limit, where it was felt that no further information could be safely taken.

Plots of exhaust gas temperature versus spark advance for each fuel and its baseline will show if there are any changes in EGT over the spark advance range as a result of the modified fuel. For each graph the significance of any differences were measured by comparison with a calculated confidence interval (CI) for the data in question. In the case of the Ricardo data, a 99% CI was used.

At spark advance points where knock was observed, LabVIEW 5.1 was used to record combustion pressure data at high speed so that it may be analysed at a later time. LabVIEW data files were converted to a Microsoft Excel format for analysis. With regular low frequency oscillations representing normal engine operation, the high frequency oscillations that occur as a result of knock detonations were counted and measured using a virtual instrument application written for the project by Frank Wu. The LabVIEW data files were then converted to a Microsoft Excel format where the proportion of knocking cycles to normal cycles was calculated. For each special fuel and its respective baseline data, a plot was made of knock proportion versus spark advance. These plots illustrate the onset of knock as spark advance is increased. By comparing the baseline and modified fuel data in these plots it is possible to assess any improvement in knock resistance as a result of the fuel.

3.4 Boroscope Inspection

After most of the Ricardo Engine runs, the inside of the combustion chamber was examined. With the spark plug removed, an ACM Inc FCB 95-A boroscope was inserted into the chamber, through the sparkplug hole. The piston crown, bore surfaces, valve surfaces and the combustion chamber were inspected. Their colour and condition was noted along with the fuel type and running conditions immediately prior to the inspection. In addition, the colour and condition of the spark plug was noted and a photograph was taken.

3.5 Spark Plug Analysis

For each different fuel type used in the Ricardo, a new sparkplug was used and the combustion deposits formed on the plug body were collected for analysis of elemental composition. In addition, a sample of exhaust pipe deposits from a trial vehicle was also taken. Six samples were examined in all.

In order to examine the samples in a Scanning Electron Microscope (SEM), the sooty spark plugs and the powder from the exhaust pipe were washed for 20 minutes in separate glass beakers with clean AR grade acetone to completely dissolve any oily residues. Each sample was then rinsed again with clean acetone to remove the oiled solvent and left to dry in air. The dried combustion deposit was removed from the spark plug face with a knife blade onto individual glass slides. A labelled SEM sample holder faced with a self-adhesive carbon pad was then pressed onto the glass slide to collect and mount the deposit powder for analysis. Each sample was sputter coated with platinum as a conductivity aid for seven minutes before being inserted into the SEM.

The SEM machine used for the analysis was a Philips XL30S FEG.

Each sample was examined visually and photographed to determine the nature of the material and any variations in appearance. Once the scan sites were selected, the samples were scanned in EDX mode to produce elemental composition data. The machine settings for each process are given in Table 2.

Table 2. SEM settings

Process	Voltage	Spot Size	Magnification
Visual & Photograph	5keV	3	1000 &10000x
Elemental Analysis	30keV	6	10000x

For each identifiably different material phase in the samples a scan graph and data table was produced by the EDX control software, showing the proportions of the component elements in each combustion deposit sample. This method allows a detection limit of 0.05 atomic percent.

3.6 Fuel Analysis

Each type of fuel used in the laboratory Ricardo engine experiments was sampled for analysis of total tin concentration. All six fuel samples were taken from the carburettor hose connection immediately before the engine was run on each fuel. The samples were collected in clean new 125mL brown glass bottles with ground glass stoppers. The bottles were filled to the top so that no air was trapped inside the bottles. The analytical chemists, Allan Aspell and Associates Ltd, performed further sample preparation, which involved shaking each sample for 30 seconds to homogenise the suspension, then dissolution and extraction of the metals with hot concentrated hydrochloric acid. Initially samples had been prepared by total digestion of the fuel sample in boiling aqua regia. The two approaches gave the same results but the second method gave a lower relative standard deviation. Dr. Allan Aspell conducted the analysis using a Perkin Elmer Optima 3100DV ICP-OES system operating in axial mode. The detection limit for tin concentration in this type of equipment is below 10 parts per billion, well below the concentrations encountered in petrol or the dilute aqueous sample.

3.7 Road Trials

All the courier companies in the Auckland region were contacted and offered the opportunity to participate in the road trials. Courier vehicles were selected for their high mileage and vigorous driving. Three courier cars and two private vehicles were secured for inclusion in the experiment. The details of each vehicle are given in Table 3.

Table 3. Road trial vehicle information

Vehicle	Year	Type	Size	Engine	Fuel System	Fuel	Canister
Toyota Corolla	1997	Car	1600cc	4 cylinder	Injection	91	PS110
Nissan Largo	1993	Van	2400cc	4 cylinder	Injection	91	Placebo
Holden Torana	1979	Car	3300cc	6 cylinder	Carburettor	96	PS115
Nissan Lucino	1995	Car	1500cc	4 cylinder	Injection	91	PS110
Honda NSR250	1993	Motorcycle	250cc	2 cyl. 2 stroke	Carburettor	96	TM-VC

The drivers’ fuel consumption logging was checked before proceeding and some minor changes were made to the information that was being gathered. Each driver then began logging baseline (pre-Fuelstar) data under their normal driving conditions and recording observations on the perceived performance of their vehicle. The drivers were to record odometer readings and fuel volume added at each refuelling point, along with comments when necessary about the general running behaviour of their vehicle. It was deemed inappropriate to demand a more stringent procedure than this because the drivers were volunteers and some were making their living from the vehicles.

3.6.1 Canister Fitting

After approximately one month baseline driving, the courier vehicles were fitted with canisters. The Holden Torana had accumulated eight months of useful data by this time and because of the lack of courier volunteers, it and the Honda NSR250 were included in the trial. Initially it was expected that three placebo canisters would be included in the trial, however the lack of vehicles and the Fuelstar sizes available dictated that only the Nissan Largo would be fitted with one of the placebo canisters pictured in Figure 3.9.

The placebo canisters were constructed from a steel pipe with 5/16” barbed hose fittings each end, going straight through an anodised aluminium tube shell that resembles a large Fuelstar unit. When connected, the fuel flowed only in the steel tube, thus exposing the fuel only to materials normally found in the fuel system, namely mild steel and zinc-plated fittings. The internal diameter of the steel tube was 8mm. The fittings were sealed with Loctite 577 hydraulic and fuel sealant. The placebo canister was fitted into the fuel line immediately before the fuel filter using the existing rubber hose upstream of the filter and a short length of Gates ¼” fuel grade hose on the down stream side. The canister was secured to the chassis of the vehicle and was visible for inspection as pictured in Figure 3.10.

In accordance with Fuelstar specifications, each vehicle’s Fuelstar canister was fitted to the engine to ensure it would be subjected to adequate vibration. The canisters were all fitted downstream of the fuel pump/filter unit and the fuel system was checked for internal filters or dust traps. Where required, new Gates 4213LC ¼” rubber fuel hose was used and all fuel lines downstream of the canister were thoroughly cleaned with MEK based carburettor cleaner, CRC Clean-R Carb.

The Toyota Corolla and Nissan Lucino (Figure 3.12) were each fitted with a PS110 as recommended by Fuelstar Ltd. In both cases, the device was mounted horizontally to the fuel system structure using the provided bracket. This orientation is identical to that shown in Figure 3.11 and other pictures provided by Fuelstar Ltd.

Figure 3.11. Manufacturer-installed Fuelstar canister (circled)

Text Box: Figure 3.12. Nissan Lucino installation

Both canisters were located downstream of the fuel pump and filter, immediately prior to the fuel injection rail such that fuel flow was in the direction of the arrow printed on the canister label. In the Corolla the existing fuel line with swaged fittings was cut in half and each end was attached to the Fuelstar with a stainless steel hose clip. The swaged fittings were then reconnected to the fuel system as before. The Lucino used standard ¼” rubber fuel hose with clips, so a length of new fuel hose was cleaned and fitted between the canister and the fuel rail. Hoses were looped to prevent kinks and restrained with cable ties where they were free to move excessively.

A PS115 was fitted to the Holden Torana in a vertical position with an upward fuel flow as suggested in the instructions provided with the product (Figure 3.13). The canister was located downstream of the fuel pump and filter, with the existing short length of steel tube connecting it directly to the carburettor. The carburettor was inspected and there were no internal filters inside.

The Honda NSR250 featured an in-tank fuel filter, which had to be removed to conform to Fuelstar directions. The inside of the tank and the fuel delivery hoses were rinsed and reassembled without the filter. The carburettors were also checked for internal filters but none were fitted. One Fuelstar TM-VC unit was then dropped into the tank when it was refilled.

With each installation, the basic theories of operation for the device were explained to the driver along with the level of overall improvement that has been claimed by the manufacturers of these devices. No information about other experimental work either current or past was presented to the courier drivers. The drivers were asked to maintain their usual driving style and that they avoid taking advantage of any power increase they may observe. The drivers remained present to observe the installation.

Once the installation was complete and all connections and safety considerations were checked, the engine was started and observed for a few minutes. At this point the driver was asked if there were any immediately noticeable differences in running characteristics. The odometer readings and an estimate of current fuel levels were noted. After a short period of static running at various throttle positions, the drivers were left to resume their normal work.

3.7 Road trial data analysis

Fuel and distance data for each vehicle was collated in Microsoft Excel along with the comments and observations made by each driver. In several cases there were events during the trial period that necessitated the truncation of some data. For each vehicle the total distance travelled and the total fuel consumed was calculated for both before and after the installation of a canister. From these totals, the total fuel economy was calculated in km/L for each half of the trial. The difference between the before and after fuel economy has been displayed in terms of a percentage change. At each refuelling point the distance travelled since the last refill (Dkm) is divided by the volume of fuel added to get a local km/L value for that point. Because there was no defined filling procedure, these values do not represent an actual fuel economy result. The average of these values was used to calculate a 90% confidence interval for the data and it was compared to the total fuel economy to assess the accuracy of the result. A 90% CI was chosen as it best showed the degree of significance of the four data sets.

For each vehicle two plots were produced. A scatter plot of distance (Dkm) versus fuel added at each refill and a plot of “local km/L” versus the number of refills were made with a polynomial trend-line fitted to the before and after data. The trend-line slopes in each scatter plot can be compared to assess any change in average fuel economy. The slope of the trend-line in the second plot indicates any ongoing change in the average fuel economy.

4.0 Results

4.1 Ricardo engine torque vs. spark advance

The Ricardo engine’s torque output was plotted against spark advance for each test fuel and its respective baseline at both 1500rpm and 2000rpm. In each plot the baseline 91-octane runs are printed in blue, while the modified fuel data is in pink. A polynomial line is fitted to the data to illustrate the overall trend. The R-squared value of each line is a measure of its accurate fit to the data and is included on the plot. The plots for the Fuelstar canister experiment are given in Figures 4.1 and 4.2. The plots of other fuel types are included in Appendix C.

To illustrate the relative effects of the different modified fuel types, an extra pair of plots have been produced which overlays all the test fuel data as points and their respective baseline polynomials. These plots are labelled Figures 4.3 and 4.4. Appendix C also contains tables showing the variation of each point from its baseline equivalent with respect to a 99% confidence interval.

4.1.1 Fuelstar

Figures 4.1 and 4.2 show the results of the Fuelstar runs. At both 1500 and 2000rpm it is clear that there was no variation worthy of note over the range of spark advance. Other than a minor loss of torque at 12º and 17º on the 1500rpm run, the difference in torque output due to the installation of the Fuelstar PS110 was less than 1% and as such, statistically insignificant.

Text Box: Figure 4.1. Torque vs. Spark advance for the Fuelstar and 91-octane baseline runs at 1500rpm

Text Box: Figure 4.2. Torque vs. Spark advance for the Fuelstar and 91-Octane baseline runs at 2000rpm

4.1.2 Granular Filtered

There was less than 1% variation in torque output between the granular filtered petrol and its baseline over most of the range. Only at 10º and 1500rpm was there a drop in torque output for the granular filtered petrol. The plots for this experiment are included in Appendix C, Figures C-3 and C-4.

4.1.3 Friction Suspension

For almost the entire range of spark advance settings, the friction suspension fuel gave a lower torque output that the baseline 91-octane run. At higher advance values where knock was observed the difference is diminished to insignificant levels. This trend is apparent at both test speeds. The plots for friction suspension fuel tests are Figures C-5 and C-6.

4.1.4 Abrasion Suspension

Only minor changes in torque are apparent in the abrasion suspension tests. A minor loss in torque from 5º to 10º advance and 2000rpm is the only variation of any significance. The1500rpm runs show barely any change at all. Plots of the abrasion suspension fuel tests are given in Appendix C.

4.1.5 96-Octane

At 2000rpm there was an across the range drop in torque for 96-octane petrol. This variation is not repeated in the 1500rpm runs. The most significant difference between the 96-octane run and the 91-octane baseline is the extended range of spark advance over which the 96-octane fuelled Ricardo engine was able to operate safely.

Figures 4.3 and 4.4 below show that even over the whole set of test runs there was only a small variation in torque output for either the baselines or the modified petrol runs. The continuous lines are fitted to the baseline data and the marked points are the test fuel data.

Figure 4.3. Torque vs. Spark advance. Combined results at 1500rpm

Figure 4.4. Torque vs. Spark advance. Combined results at 2000rpm

4.2 Ricardo knock proportion vs. spark advance

The results presented here illustrate the change in knock behaviour of the Ricardo engine as a result of the different fuel types. Knock proportion is a measure of how many combustion cycles in the sample of 222 cycles exceeded the knock threshold criteria. A plot has been made for each experiment at each engine speed showing the knock proportion relative to the spark advance. A trend line is fitted to the test fuel data and its baseline, which highlights any differences due to the modified fuel. A full set of plots for each experiment is printed in Appendix D.

4.2.1 Fuelstar

Figure 4.5 shows the knock proportion for the Fuelstar experiment at 1500rpm. There is very little difference between the Fuelstar and the baseline runs. The 2000rpm run gave similar results and its plot is available in the appendices. This test showed the least variation of all the knock data.

Figure 4.5. Knock proportion vs. Spark Advance for Fuelstar and

91-octane baseline at 1500rpm

4.2.2 Granular Filtered

The granular filtered petrol result varied slightly from its baseline, but not to the extent that it can be attributed to the treatment of the fuel. The plots for this fuel in Appendix D lack baseline data at the highest spark advance points as a result of a recording error.

4.2.3 Friction Suspension

This experiment gave the biggest variation of all the modified fuel types. At all the points shown in Figure 4.6 the friction suspension fuel has a lower knock proportion that the baseline. Despite these measured results, observation of the oscilloscope and audible knock suggest no improvement in knock resistance. The same trends are apparent in the 2000rpm data presented in the appendix.

Figure 4.6. Knock proportion vs. Spark advance for friction

suspension and 91-octane baseline at 1500rpm

4.2.4 Abrasion Suspension

The abrasion suspension exhibits a slight reduction in knock proportion relative to the baseline values. At 1500rpm the results are inconclusive and incomplete. The 1500rpm run was interrupted due to a shortage of treated fuel and a misjudgement of the test duration.

4.2.5 96-Octane

To put the variations into perspective, a plot of knock proportion for 96-octane fuel has also been produced. The Ricardo engine was not pushed into as severe operating conditions for this run, so the percentage of knocking cycles is lower than the other data. Even at these lower levels the difference between the two fuel grades is dramatic. The 91-octane fuel was heard to knock heavily at a point where the 96-octane fuel was still able to perform perfectly well. This observation is illustrated in the graph, Figure 4.7. The extent of the difference is obviously far greater than that experienced in any of the other tests. The 2000rpm plot given in Appendix D shows the same degree of difference between the two fuel grades.

Figure 4.7. Knock proportion vs. Spark advance for 96-octane and

91-octane baseline at 1500rpm

4.3 Observations of oscilloscope and audible knock intensity.

During every Ricardo engine experiment notes were made regarding the onset and intensity of knock as heard and seen on the oscilloscope in real time. When knock was observed occasionally on the oscilloscope, but still inaudible a label of MK or “minor knock” was applied to that setting. If the knock was constantly audible and visible on the oscilloscope pressure trace, a label of KL or “knock limit” was applied as this was estimated to be severe enough to cause significant damage to an ordinary automotive engine. In every case the knock limit was identified at the same spark advance setting in both the test fuel run and its respective baseline. Table 4 below shows the audible and visible knock results.

Table 4. Visible and audible knock limits

Fuel type	MK (1500)	KL (1500)	MK (2000)	KL (2000)
91-octane	11	12	15	18
Fuelstar	11	12	15	18
91-octane	12	14	16	19
96-octane	16	19	20	22
91-octane	13	15	18	20
Granular	14	16	19	23
91-octane	14	16	19	23
Friction	14	16	19	23
91-octane	13	16	18	22
Abrasion	13	16	19	21
91-octane	14	16	18	23
Fuelstar	14	16	17	23

Two cases show a slight observed improvement in knock behaviour. The 96-octane test resulted in a clearly defined change in minor knock and in knock limit. The granular filtered fuel gave a smaller improvement in each defined limit.

4.4 Pre-ignition

During an early trial of a hot spark plug and during the 2000rpm run with Friction suspension petrol occasional pre-ignition was observed. The overlapped waveform plots from the LabVIEW software showed that pre-ignition was occurring regularly at the higher ignition advance settings. The separate, rising curve in Figure 2.1 represents pre-ignition.

4.5 Exhaust gas temperature

Exhaust gas temperatures (EGT) were monitored and recorded for every data point during Ricardo engine runs. Data from the 2000rpm runs was compromised, probably by electromagnetic interference. The proximity of the thermocouple and its wires to the ignition system is believed to have caused these anomalies. The EGT data from 1500rpm runs was fairly consistent, but the slope of the plotted data is not in the direction expected. The results show that EGT does not change significantly as a result of adding tin alloy of any form to the fuel. In every case a stable maximum EGT of approximately 568ºC is reached. Figure 4.8 shows the result for Fuelstar at 1500rpm. The complete set of EGT data is printed in Appendix E.

Figure 4.8. EGT vs. Spark advance for Fuelstar and 91-octane

baseline at 1500rpm

4.6 Atmospheric data

For every Ricardo engine run the atmospheric conditions at the air filter inlet were recorded mid-run for comparison. The temperature, pressure and humidity data is summarised in Table K-1 of Appendix K. Over the duration of Ricardo engine testing there was some variation in conditions, however between test fuel and baseline of each type, there were no significant variations that could account for any performance differences or lack thereof. The 96-octane experiment experience the largest variation, but this was not sufficient to have seriously influenced the changes observed.

4.7 Boroscope inspection and spark plug appearance

Observation of the spark plugs after each Ricardo run showed no obvious changes in the nature of the combustion deposits on the surfaces of the spark plug. In each case, the body perimeter was covered evenly with dry black soot. The centre electrode tip and the earth electrode tip of each plug were relatively clean and grey in colour each time. With the exception of the 96 octane petrol, all fuels left the sparkplug insulator with a dark grey deposit coating and a small clean patch near the electrode. The plug used in 96 octane runs had a relatively clean ceramic insulator after the final run. The appearance of spark plugs did not seem to be greatly affected by the running speed immediately before inspection. Figure 4.9 shows a clean plug and Figure 4.10 is the Fuelstar plug. Pictures of all the spark plugs are printed in Appendix H.

Text Box: Figure 4.9. Clean sparkplug

Text Box: Figure 4.10. Fuelstar sparkplug

Boroscope examination of the combustion chamber showed some changes in appearance, but no observable trends. There appeared to be a tendency for more combustion deposits after running at 2000 rpm than at 1500rpm. This difference was not observed between treated and untreated fuel runs. Generally the combustion chamber and piston crown surfaces were coated with slightly oily looking black deposits. The inlet valve consistently had dry dusty looking brown/grey deposits over the face and stem side, with the seat being fairly clean and shiny. The exhaust valve began with the same brown/grey appearance as the inlet valve, but after the first Fuelstar run, its face was covered with an even white/grey colour. After the next run on 91-octane petrol the colour reverted to dark grey, but the white colour re-appeared after further 91-octane running. For the remainder of operation, the exhaust valve colour remained more or less light grey. After the final Fuelstar run there appeared to be some flakes forming on the exhaust valve seat. It is uncertain whether this was a new development or simply the result of valve rotation exposing the other side of the seat to inspection.

4.8 Spark plug analysis

Results of the SEM analysis of the spark plug deposits are summarised in Table 5. The limit of resolution for the equipment used is 0.05%.

Table 5. Elemental Analysis of Combustion Deposits (Atomic %)

All the deposits consisted mostly of carbon with a smaller proportion of oxygen and some common metals. Of particular note are the low levels of tin (Sn) and antimony (Sb) detected. Only the fuel with a relatively high concentration of abraded tin alloy particles produced deposits with significant quantities of tin or antimony. An unexpected amount of calcium (Ca) was apparent in several samples. The exhaust pipe sample from the Holden Torana shows an unusually large proportion of silicon (Si), while the Fuelstar samples are high in all metals except tin and antimony, with lead (Pb) and zinc being especially high. Several glass fibres were observed in the sample under the SEM.

4.9 Tin concentration analysis

The Certificate of Analysis produced by Allan Aspell and Associates is included in Appendix B. The results of this analysis are given below in Table 6. The concentration values (mg/kg) are equivalent to parts per million (ppm) and are relative to the original petrol samples. The detection limit for these results is below 0.01ppm.

Table 6. Total tin concentration of petrol samples

Sample	Treatment	Tin (mg/kg)
1. 91 Octane Petrol	None	0.13
2. Fuelstar	One pass through a vertical PS110 vibrating at 28Hz, 0.8mm amplitude	3.1
3. Granular Unfiltered	20 hours agitation with 16.7g of tin alloy granules.	0.16
4. Granular Filtered	As Sample 3 but filtered twice through grade 4 and 5 filter paper.	0.16
5. Abrasion Suspension	0.3g of tin alloy added by abrasive wear.	96
6. Friction Suspension	0.16g of tin alloy added by friction wear.	15

4.10 Road trials

Road trial results are presented in terms of “fuel economy” with units of km/L. For each vehicle the fuel economy was calculated as a total value and an average value. The total fuel economy for the before and after installation sections is derived by dividing the total distance travelled in that section by the total fuel used. The average fuel economy is calculated as the average of the values calculated for each refuelling stop. Because of the large variation in these values, this is the less reliable Figure. It is most useful as a gauge of the accuracy of the test overall.

A Honda NSR250 motorcycle was initially included in the trial, but shortly after the installation of a Fuelstar device and removal of the in-tank fuel filter this vehicle began to suffer sharply diminished performance at mid to low speed and the engine would no longer idle for longer than a few second. As a result of this problem the vehicle was withdrawn from further participation in the trials.

Table 7 summarises the calculated fuel economy data for each vehicle and presents the percentage change along with a 90% confidence interval.

Table 7. Fuel consumption results summary

		Total			Average
Vehicle	Pre-FS	Post-FS	Difference	Pre-FS	Post-FS	Difference	90% CI
	(km/L)	(km/L)	(%)	(km/L)	(km/L)	(%)
Corolla	11.948	12.031	0.693%	11.973	12.061	0.734%	1.636%
Largo	8.071	7.859	-2.626%	8.057	7.812	-3.043%	3.724%
Torana	7.528	7.444	-1.128%	7.809	7.387	-5.392%	32.134%
Lucino	12.376	12.170	-1.665%	12.518	12.296	-1.777%	7.056%

The Fuelstar equipped Toyota Corolla was the only vehicle to experience any kind of fuel saving. However the difference is extremely small and lies well within the 90% confidence interval. The other two Fuelstar vehicles, the Holden Torana and the Nissan Lucino showed small decreases in fuel economy after installation of their canisters. Once again the changes are well within the confidence interval suggesting that they are of no significance. The Nissan Largo was fitted with the placebo canister. It also showed a small, statistically insignificant decrease. Scatter plots of the data confirming these values are printed in Appendix I.

Figures 4.11 to 4.14 below show how the average fuel economy changed with time. Time is given in terms of refuelling stops in this case. Data from before the installation of a canister is in blue, while the data from after the installation is printed in pink. The trend line fitted to the data loosely translates as the average fuel economy.

Figure 4.11. Toyota Corolla. Change in fuel economy after Fuelstar installation

Figure 4.12. Nissan Largo. Change in fuel economy after placebo installation

Figure 4.13. Holden Torana. Change in fuel economy after Fuelstar installation

Figure 4.14. Nissan Lucino. Change in fuel economy after Fuelstar installation

The trend lines, in the case of the Corolla (Figure 4.11) and the Lucino (Figure 4.14) confirm that there is no difference in fuel economy over time and suggest that there is no tendency for a change. Figure 4.13 shows that the Torana’s fuel economy had declined during the initial period. The Nissan Largo plot (Figure 4.12) suggests that despite the small measured decrease in fuel economy, the tendency was for an increase as time passed.

Each driver’s comments regarding any change in performance as a result of fitting the canister are summarised by the following:

· Toyota Corolla driver: “5-10% more power; better starting; smoother running”.

· Nissan Lucino driver: “More power; a fuel saving; hardly stalls; generally runs better”.

· Nissan Largo driver: “Starts a bit better, its not making it worse”.

· Holden Torana driver: “Hesitant at just below 2000rpm, no improvement in pinging”.

5.0 Discussion

5.1 Tin concentration

The measured tin concentration of each fuel follows quite closely to the amount of material that was added during preparation. Granular filtered and unfiltered samples were both barely above the 91-octane control sample and were both the same, suggesting that the tin was mostly in a form to minute to be retained by the filter paper. The Fuelstar canister was able to provide 3.1ppm to the fuel sample with one vibrating pass. This is consistent with the experiment performed in Australia [9], but a French laboratory managed to achieve ten times this concentration under similar conditions [10]. An estimate of the maximum possible tin attrition was made by considering the average fuel consumption of a small car, the mass of metal available in the canister and the 500,000km warranty period for the Fuelstar product. This indicated that 4 -7ppm was sustainable under these conditions. The friction suspension sample was created by a vigorous simulation of the wear conditions inside a vibrating canister and it shows a correspondingly higher concentration of tin, 15ppm. This value is about five times the Fuelstar result. The mass of tin removed from the rubbing mechanism would have given a concentration of 56ppm had it all dissolved. The difference indicates the quantity of large particles that quickly fell out of suspension and sank to the bottom of the vessel during preparation. Microscopic examination of these particles showed that they ranged from 2-20 microns with some up to 60 microns across. The smallest visible particles were 500nm, which suggests that those remaining in suspension were of this order and smaller. The highest tin concentration occurred in the abrasion sample as expected. The goal was to produce 128ppm, which is the same concentration that was specified for leaded petrol in many countries. The sample was measured at 96ppm; some 30 times more than was available from the Fuelstar canister. The particles were measured under an optical microscope and averaged 4 microns across. They were very regular in size and shape with the range extending from 2 –10 microns.

5.2 Torque

Over the course of Ricardo engine experiment there was very little significant variation even when considering the most extreme results collected. When comparing the individual fuels with their respective baseline runs, there is a universal lack of improvement in torque output for fuels treated with any form of the Fuelstar tin alloy. In most cases any change was a negative one, but these changes were so small that they rarely crossed the bounds of a 99% confidence interval. This suggests with 99% confidence, that there were no effects on the Ricardo engine’s torque output as a result of installing the Fuelstar canister or treating the petrol with tin alloy.

Between the early Ricardo engine data and the later experiments there was a change in the limits to which the engine was operated. The knock limits for the later runs were 2 –3 degrees more advanced in some cases. It has been suggested that the occasional input of tin into the system during subsequent experiments had a lasting effect on performance, which was not quenched by the baseline runs. This is not believed to have been the case because of the fact that some of the preliminary runs of the Ricardo engine showed knock limits of the same order as the last few experiments. A large part of this difference is the initial reluctance of the operators to run the engine in severe knock conditions. The low level of knock shown in the pressure trace analysis of the first few runs confirms that this early concern was unnecessary. Minor changes in weather conditions may have also contributed a small effect. The biggest effect is likely to have been the idle speed experiments that were being performed concurrently with the early experiments. This sustained idle operation would have left relatively thick combustion deposits on valves and piston crown which would take some time to be burnt off with more vigorous running.

Temperature, pressure and humidity changes over the course of the experiments appear not to have greatly influenced the results. The greatest variation occurred during the 96-octane runs when it rained outside for a short time. This perhaps accounts for the diminished torque of the 2000rpm result.

It may be concluded from these results that Fuelstar and similar tin based fuel catalyst products are unable to improve the torque output of a petrol engine. The variations and influencing factors encountered are small and can be accounted for, suggesting that this result is entirely valid. As these results give no indication of an effect, it is not possible to make assumptions as to the mechanism by which this technology is supposed to work.

5.3 Knock and pre-ignition

The 96-octane results for knock proportion are useful as a benchmark of what degree of improvement is available with the higher grade of petrol. The regularly made claim that tin catalyst products will allow a vehicle designed to run on 96-octane to safely run on 91 octane, suggests that the same order of improvement should be apparent.

The results of the Fuelstar experiment show that this is clearly not the case. The Fuelstar PS110 results show that there was effectively no improvement in knock behaviour of the Ricardo engine as a result of the canister.

Granular filtered and abrasion suspension fuel types also showed no useful improvement in knock although the reduced quantity of data for the abrasion suspension fuel leaves this result uncertain. The granular filtered fuel results suggest that a catalytic reaction on the tin alloy surface is not sufficient to aid knock resistance. It also appears that despite the large quantity of tin in the abrasion suspension fuel, the mass of metal transferred to the combustion chamber does not affect knock.

The experiment that did give a positive result was for the friction suspension fuel type. The plot of knock proportion for this fuel (Figure 4.6) indicates that knock was delayed by 1 or 2 degrees. The 96-octane fuel by comparison was able to avoid knocking for about 5 degrees. The main difference between the friction suspension fuel and the others was that it probably contained a far greater proportion of colloidal particles that measure below 100nm in diameter. The massively increased surface area per mass of this type of particle could explain the difference in performance. A moderate increase in relative humidity and a drop in temperature between the baselines and the test runs would have also influenced this result slightly. This improvement in knock resistance was not observed audibly or on the oscilloscope as the run proceeded. These results suggest no improvement, although the magnitude of change shown on the plot may be too small to have been noticeable by direct observation.

During the 2000rpm run with the friction suspension fuel an occasional pre-ignition event was observed. Analysis of the pressure trace data showed that pre-ignition occurred about 4 times in the whole data set which confirmed the regularly audible deep thump from the engine. The only other time pre-ignition occurred was during a preliminary test run using a very hot spark plug. This phenomenon was probably caused by a hot spot on the plug or combustion chamber surface. A combustion deposit flake or the larger tin particles in this fuel could provide such a hot spot.

For the most part these results do not shed much light on the mechanism by which knock is influenced by tin alloys. The only deduction that can be made in this respect is that surface area is likely to be of more consequence than material mass.

The data collection techniques used for pressure and consequently knock information are highly accurate, but the erratic nature of the phenomenon means that a large sample must be taken. The audible, oscilloscope and pressure trace assessments of knock behaviour have a fairly good correlation to each other, but the variations indicate that a larger pressure data sample would result in a much more accurate picture of knock behaviour.

These results suggest that installation of a Fuelstar or similar tin based catalyst does not cause any improvement in knock behaviour. It appears that if the friction suspension result does represent an actual benefit, it is likely to be the result of high surface area tin alloy particles entering the combustion chamber, however there are sufficient external influences to cast some doubt on the actual influence of the fuel.

5.4 Exhaust gas temperature

Exhaust gas temperature records from the Ricardo engine indicate that there was very little change between the results. In most cases the temperature ranged from 400ºC at zero advance to 568ºC at the peak. The 1500rpm plots shown in Appendix E differ substantially from the expected form in that the temperature rises with increasing advance rather than dropping in a fairly linear fashion with increasing ignition advance. EGT data from the 2000rpm runs followed a wildly erroneous path and it was consequently not plotted.

Two factors are known to have influenced these results. The K3 type thermo couple fitted to the exhaust manifold was not calibrated beyond 500ºC and the leads from the thermocouple to the charge amplifier passed within a few centimeters of the ignition module and coil. The electromagnetic interference from the ignition system is probably the main cause of the problem. Consequently, the EGT data is of limited usefulness. On the assumption that the variation of the data is constant throughout the experiments, it may be tentatively concluded that there was no significant difference in EGT as a result of the Fuelstar canister or the tin alloy modified fuels. A drop in EGT is occasionally claimed as a benefit of tin alloy fuel catalysts, but this claim is not confirmed by these results.

5.5 Combustion deposits

A regular comment amongst satisfied Fuelstar users is that they see a white deposit on the spark plugs, piston crown or inside the exhaust pipe. This is generally assumed to be an oxide of tin. During the road trials of the Holden Torana, the previously black exhaust pipe soot was seen to be a dark grey colour for a short time. The SEM analysis showed that this sample did not contain any amount of tin that could have caused this colour. High humidity on that day may have caused a “blushing” effect where moisture is adsorbed onto the soot surface. The Torana sample had very high silicon content which was quickly explained when a short strand of glass fibre, presumably from the muffler packing, was seen buried in the sample soot.

Boroscope examination of the Ricardo engine exhaust valve did reveal a dramatic change in its colour after the first use of a Fuelstar canister. This change in colour did not however, remain linked to the use of tin bearing fuels. It was seen to both diminish and return during subsequent 91-octane fuel runs without any further tin exposure. This remains an interesting and unexplained event and should be noted for future investigations.

Spark plug deposits were identical in appearance for all fuel experiments. The SEM elemental analysis results show that even the very high quantities of tin in the abrasion suspension deposit could not produce a grey or white colour. This was the only combustion deposit sample that showed a significant amount of tin content. The fact that new plugs were used for each sample and that the results are in terms of atomic percentage, suggests that a longer running time would not have changed the composition of the deposit.

Over the course of the experiments the total bulk of combustion deposits inside the combustion chamber appeared to gradually diminish. This could be a cumulative effect of the tin content, but it is more likely to be the result of the vigorous running the engine experienced after the extended idling it did in the previous project it was used for.

The only correlation between the tin content of the combustion deposits and the tin concentration of the fuels used is shown by the abrasion sample. All the other deposits registered below the limit of detection for the SEM. Other results of note include the high calcium level in most of the results, probably from a cooling water leak into the combustion chamber leaving calcinous deposits behind. The fact that calcium is present in the Torana sample too, but not is some of the others indicates that it could be from contamination of the sample during preparation. The lead and zinc levels in the Fuelstar sample were also unexpected. Zinc is present in the carburettor, engine oil, spark plug plating and on internal components of the canister, so it’s presence is easily accounted for. Lead on the other hand is somewhat of a mystery. It is known that the tin alloy contains a small component of lead, but not enough to explain the high level found. The other possibility is that some residue remained in the engine or fuel system from the days when the engine was supplied with leaded fuel. In this case it is assumed that lead would have appeared more significantly in the other samples too.

Possibly the most interesting observation is that the 91-octane sample has such a low oxygen peak that it was not measured. It is estimated from Figure G-1 that this sample contained less that 10% of the oxygen present in some samples. This indicates that there is some difference in the oxidation of combustion products when tin alloy is introduced to the system.

5.6 Fuel consumption

For all four vehicles tested there was no significant change in fuel economy during the trial. Three of the vehicles experienced minor drops and one a tiny increase, but in each case the difference was well within the 90% confidence interval, and mostly within a 99% confidence interval. In addition to the lack of total and average change, there was also a lack of trend to change as shown by the relatively flat curves in figures 4.11, 4.13 and 4.14. The exception to this was the Nissan Largo, which was fitted with a placebo canister.

The degree of variability of the data collected for these vehicles means that there would need to be a fairly significant change in fuel economy for it to be considered a genuine result. Considering this, it is useful to compare the results with the 10% improvements often claimed by manufacturers. Even considering the margin of error, there is no possibility that the canisters fitted to the trial vehicles were producing a 10% improvement. Furthermore, it can be concluded with 90% confidence that there was no significant improvement in fuel economy as a result of fitting Fuelstar canisters.

The comments made by the Nissan Lucino drivers are in sharp contrast to the measured results of the vehicle. He was adamant that there was a fuel saving along with the other benefits he noted. Improvements in power and running conditions cannot be verified, as these parameters were not measured. It was initially expected that the vehicle fitted with the placebo canister would experience the same improvements as the other vehicles, but as there was no improvement in any vehicle, an assessment of the placebo effect is difficult. A much larger sample size is required to get useful information about the influence of psychological factors. It is interesting to note however, that the two drivers who gave highly positive reports of the Fuelstar’s benefits were the ones who were given the most detailed explanation of the manufacturers expectations for improvement.

The results of all the experiments performed show that there is no evidence that installation of a Fuelstar canister or a similar tin based fuel catalyst product can significantly improve torque output, knock resistance, fuel economy or reduce exhaust gas temperature, particularly over a short time frame. There remains the possibility that improvements can be made over extended periods as a result of combustion deposit control.

The mechanism by which these products work is still unclear, although there is some evidence to suggest that the placebo effect may have an influence and that knock behaviour may be affected by colloid sized tin alloy particles. The variability of the data collected for this experiment shows that very strict data collection procedures are required to obtain an accurate result, illustrating that usefulness of much of the anecdotal evidence of fuel savings is doubtful.

More detailed research into combustion deposit control, the effect of high surface area particles on knock and psychological effects such as the placebo effect may be able to shed more light on these areas of the subject that are still uncertain.

6.0 Conclusions

Fuelstar had no significant effect on the Ricardo engine’s torque output or knock limit in the laboratory.
Fuelstar had no significant effect on the fuel economy of the road trial vehicles.
Fuelstar does release small amounts of tin into the fuel stream with sufficient vibration.
Tin concentrations in combustion deposits reflect the tin concentration of the fuel used.
The use of Fuelstar and the use of fuels that contained added tin alloy did no measurably affect deposit formation over a short timeframe.
Fuels containing tin particles of varying quantities and concentrations did not benefit Ricardo engine performance significantly.
Fuel exposed to a high surface area of tin alloy for an extended time did not significantly benefit Ricardo engine performance.
The use of a single placebo canister in the road trial did not yield useful results.
Regulated road trials with a large number of vehicles and long duration Octane Requirement Increase experiments on a laboratory engine are required.

Word Count: 14,661

7.0 Further Work

While the results of this project and several previous investigations show that tin based fuel catalyst products do not significantly benefit engine performance, there is potential for long-term effects, which take some time to develop. It has been shown that a vibrating Fuelstar canister is capable of releasing tin into the fuel stream, and this gradual addition of material to the combustion chamber may with time provide some beneficial effect.

There are three areas where further work could answer the remaining questions about the effectiveness of tin based catalysts:

1. Measurement and analysis of vibration in average road driven vehicles compared to the vibration of the catalyst canisters fitted to vehicles which have reportedly improved performance as a result of tin catalyst installation.

2. Long term road trials lasting six months or more with a large number of vehicles. Several anonymous placebo canisters should be included in selected vehicles to assess the psychological effects of drivers’ expectations. Some vehicles should start with a canister fitted and then have it removed, while others should have the canister fitted for the last half of the trial. This will enable assessment of transitional effects.

3. An Octane Requirement Increase test of untreated fuel over 250 hours running time on a laboratory engine followed by the same test with a tin catalyst product installed. This experiment will show if there are any long-term benefits as a result of deposit formation control.

4. An investigation of heavy metal emissions from vehicles fitted with tin alloy fuel catalyst products.

8.0 References

1. Kovarik, B. Charles F. Kettering and the 1921 Discovery of Tetra Ethyl Lead in the Context of Technological Alterations. (1999). [online]. Viewed 27/4/2002 www.runet.edu/~wkovarik/papers/kettering.html

2. Heywood, J.B. (1988) Internal Combustion Engine Fundamentals.

McGraw-Hill, Inc. USA

3. Weaver, C S. (1986) Effects of Low Lead and Unleaded Fuels on Gasoline Engines.

SAE Technical Paper Series 860090, Society of Automotive Engineers, USA, p3.

Schoonveld, G A., Riley R K., Thomas S P. & Schiff S. (1986) Exhaust Valve Recession with Low-Lead Gasolines. SAE Technical Paper Series, 861550, Society of Automotive Engineers, USA pp3-4.

4. International Tin Research Institute, UK [online] viewed 30/4/2002

http://www.itri.co.uk/fcats.htm

5. http://fp.rdg.ac.uk/wkc1/diy/car/msg00338.html

6. Prozone Fuelsaver. [Online] viewed 22/4/2002

http://www.fuelsaver.co.uk/howitworks.htm

7. Fuelstar International Ltd, [online] viewed 15/5/2002

http://www.fuelstar.com/performance.html

http://www.fuelstar.com/fuel_savings.html

http://www.fuelstar.com/unleaded.html

http://www.fuelstar.com/emissions.html

8. Office of Fair Trading, UK. [online] released: 24/10/2000, viewed: 30/4/2002

http://www.oft.gov.uk/news/press+releases/2000/pn+42-00.htm

9. Wilkinson, R. A (1998) Fuelstar Fuel Catalyst. Australian Fuel Developments, McRae Victoria

10. Fuelstar Concentration Tests (translation) (1999) UTAC Direction Technique, PROCES-VERBAL No 99/08179. Prepared for:T5 Developpement, St Genis Laval

11 Lewis R.J (1997) Hazardous Chemicals Desk Reference. Fourth edition. International Thomson Publishing Company, New York. pp76-78,1147-1148

12. PCImag.com. Tributyltin Antifouling [Online] viewed 22/9/2002 http://www.pcimag.com/CDA/ArticleInformation/features/BNP

__Features__Item/0,1846,5695,00.html

13. Ferguson C. R. (1986) Internal Combustion Engines: Applied Thermosciences. John Wiley & Sons, New York. p442

14. Sprott, T J. (2001) Desktop Study: Fuelstar Combustion Catalysts. [Online] http://www.fuelstar.com/data/Desktop%20Study%20Fuelstar%2015-12-01.pdf, p.4

15. Pearce, J. (1991) Tin Alloy Based Hydrocarbon Catalysts, Report 4: The Chemistry of Interaction of Tin Alloys with Hydrocarbons.

International Tin Research Institute, UK.

16. From correspondence with Dr. Matthew E. Roser, Dartmouth College Psychology Dept. Hanover, NH USA 12/5/2002

17. Sprott, T J (2001) Desktop Study: Fuelstar Combustion Catalysts. [Online] http://www.fuelstar.com/data/Desktop%20Study%20Fuelstar%2015-12-01.pdf

18. EFRU, (2002) Fuel Economy Testing of Fuelstar Device.

Auckland UniServices.

EFRU, (1998) Effects of Fuelstar Device on Exhaust Emissions, Fuel Consumption, Power Output and Knock. Auckland UniServices.

19. Watson, H. (1997) Valve Seat Recession and Knock Limited Spark Ignition Timing Tests, Part 1: A Fuelstar Device. New Zealand Automobile Association.

20. The Federation of British Historic Vehicle Clubs & MIRA, Fuel Additive Tests [Online].viewed 15/2/2002 http://ourworld.cs.com/motordata/page24.html

21. Cotton, S. Tetra Ethyl Lead & MTBE. [Online] viewed 12/4/2002

http://www.chm.bris.ac.uk/motm/leadtet/leadh.htm

22. McArragher, J S; Clarke, L J & Paesler, H. (1993) Prevention of Valve Seat Recession in European Models. Co-ordinating European Council

23. Weaver, C S. (1986) Effects of Low Lead and Unleaded Fuels on Gasoline Engines.

SAE Technical Paper Series 860090, Society of Automotive Engineers, USA, pp7-10.

24. Schoonveld, G A., Riley R K., Thomas S P. & Schiff S. (1986) Exhaust Valve Recession with Low-Lead Gasolines. SAE Technical Paper Series, 861550, Society of Automotive Engineers, USA p2.

25. Fenton J. Ed. (1986) Gasoline Engine Analysis. Cambridge University Press, UK.

9.0 Bibliography

· Taylor D. F. (1966) The Internal Combustion Engine in Theory & Practice. Vol 1. The M.I.T Press, Cambridge Massachusetts

· Plint M. and Martyr A. (1995) Engine Testing: Theory and Practice, Second edition. Butterworth-Heinemann, Oxford.

· Ebert L. B. (1985) Chemistry of Engine Combustion Deposits, Plenum Press, New York

PROJECT IN MECHANICAL ENGINEERING

Acknowledgements

Table of Contents

Glossary