Eric Anderson, Henry Wandrie, John Celmins,

Joseph Bouboulis, Colin Botha, Jason Dalton,

Denis Klug, Timothy Kish, Slaven Sljivar,

GMI Engineering & Management Institute

The 1997 Propane Vehicle Challenge offers students a chance to incorporate classroom methods into the very real world of vehicle design. The challenge is to develop a LPG powered vehicle which performs better than a gasoline fueled vehicle in the areas of performance, economy, emissions, and initial cost.

Propane as an automobile fuel carries an 84 year heritage, with the first propane based vehicle being developed in 1913. Currently, LPG supplies 3-4% of our total energy needs [1].

LPG, which has propane as its primary constituent (see table 1), has the advantages of producing less emissions, offering more energy per unit mass, and a higher octane rating than gasoline. In addition, through the use of LPG we can continue to work away from dependence on foreign companies to supply domestic energy.

LPG also has the advantage of already having an infrastructure in place to support its implementation not only as an alternative fuel, but as a primary fuel. This infrastructure is represented by over 9,000 bulk/storage distribution points, and 25,000 retail outlets [1]. With the advent of legislation for cleaner burning vehicles, LPG's low emission properties present it as an attractive solution for domestic energy needs.

There are several alternative methods for propane conversion. These methods include: gaseous carburation, gaseous injection and liquid injection. Each method has advantages and disadvantages.

While the gaseous carburation system has the advantage of availability of its technology, it is not compatible with today's electronically port fuel injected (EFI) engines. Because of this incompatibility, gaseous carburation on a modern EFI engine can be dangerous. When the vehicle is no longer running, the stoichiometric air/fuel mixture is allowed to stagnate in the intake manifold. The heat provided by this heat soak can induce combustion in the manifold, producing catastrophic failure.

A gaseous injection system is a relatively simple modification to implement. This system has the dual advantages that it requires less modifications than that of gaseous carburation or liquid injection for proper fuel delivery, and that it does not pose the threat of the fuel changing states during prolonged operation. These advantages, however, are offset by a loss in performance. Because injected propane vapor can require 4% to 15% of the available intake passage volume, space which would normally be occupied by air necessary to maintain the power of the engine, the gaseous injection system can have a power loss of as much as 4% when compared to its gasoline counterpart [3]. The only way to overcome this drawback is either to directly inject the fuel into the combustion chamber, increase the compression ratio, or use forced induction (turbocharge or supercharge) to increase the amount of air supplied to the combustion chamber [3]. Both of these methods would add complications to the originally simple modification.

The third conversion method is liquid injection. This method does have some design drawbacks, but these drawbacks are offset by an improvement in performance. Because LPG does not occupy a significant amount of volume in the intake, the performance will not be reduced. Furthermore, the LPG will vaporize as it mixes with the air, absorbing energy from the air, lowering the air's temperature, and ultimately increasing its density, again increasing performance [3]. This advantage over a gaseous system eliminates the need for forced induction in order to maintain performance. The single greatest drawback of a liquid injection system are problems associated with maintaining the propane in a liquid state up to the injectors, given the operating temperature of the engine.

Using data acquired from the '96 Propane Vehicle Challenge, and subsequent operation of our vehicle, the GMI team developed a design strategy targeted at overcoming last years difficulties, and optimization of LPG as our fuel. After developing a list of total possible modifications, and eliminating those which did not fit our design goals as outlined in our abstract, we chose to make those changes which offered the greatest return on investment for the manufacturer, customer, and the '97 Propane Vehicle Challenge.

Three modifications were made to the engine. First the compression ratio was increased to take full advantage of the fuel's high octane number. The second modification included insulating the intake and exhaust manifolds from the engine to reduce heat transfer which in turn increases the performance and decreases the emissions. The final modification utilized a teflon coating to minimize frictional losses in the intake flow field.

The most significant advantage of an increased compression ratio is a better utilization of the knock resistance of LPG (see table 2), which results in an increase in engine output, as well as an increase in fuel efficiency. However, an increase in compression ratio does present the drawback of increased NOx emissions. The final engine design was chosen to balance performance and emissions.

From an analysis of the air Otto Cycle it is apparent that increasing the compression ratio (R_c) will produce a better performance, however, this compression ratio increase will also result in higher cylinder temperatures. As the temperature rises so does NOx emissions. Based upon LPG's octane rating (an indication of it's resistance to knock, see table 2), last year's competition results, and our emissions goals, we increased the compression ratio of the engine to 11.1:1.

To achieve this increase in the compression ratio new pistons were designed to reduce the clearance volume. The pistons which were used, were manufactured by Ross Racing Pistons from a design which they had already produced for the 1996 Propane Vehicle Challenge. This reduced the lead time and cost associated with producing the pistons.

Because of the high thermal conductivity of the engine and manifold materials, heat transfers readily from the engine to the intake manifold. This results in an elevated manifold temperature which in turn results in an elevation in the incoming air's temperature. This temperature change will change the density of the air as shown in Figure 1. This decrease in the density of the air entering the combustion chamber ultimately lowers the volumetric efficiency of the engine.

Figure 1: Relationship between density and temperature for air.

Two steps were taken to reduce the heat transfer from the engine to the manifold. The first step was to install a phenolic spacer, which has a very low thermal conductivity, between the engine and the intake manifold. The second step was to apply a ceramic coating to the intake manifold.

By using ceramic coating on both the intake and exhaust manifolds, the incoming air was insulated from heat addition while the exhaust was insulated from heat losses. This reduction in the heat loss from the exhaust caused a decrease in the amount of time for the catalyst to reach maximum efficiency. During startup, when the catalytic converter is cold, many unwanted emissions escape into the atmosphere. The catalytic converter, in its most basic form, performs a chemical reaction which converts unwanted pollutants into other chemicals which are not detrimental to the environment. By insulating the exhaust runners the catalyst light off time is significantly reduced.

The final modifications considered the affects of air flow on the performance of the engine. Through research it was found that the addition of a teflon coating on the inside of the throttle body on Chrysler's 3.3 liter gasoline powered engine results in 28% more air than stock at 30% open throttle due to the reduced friction between the throttle body and the air [4]. The GMI throttle body was coated by Polydyn & Intake, of Texas, as well as having the throttle plate profiled by Peter's Engineering for the smoothest possible flow. Air flow restrictions were also minimized by port matching throughout the intake system. This smoothes out overlaps in material caused by standard production tolerancing.

The goal in developing the exhaust system was to increase flow (and therefore performance) while reducing emissions. By decreasing exhaust back pressure better scavenging, a measure of the amount of burnt material left in the cylinder from the previous cycle, can be achieve.

The GMI exhaust system uses a 63.5 mm mandrel bent stainless steel exhaust system manufactured to specification by Watson Engineering, Taylor, Michigan. This change increases the cross sectional flow area by 56%. Due to tank location and configuration, it was necessary to reroute the exhaust from its stock location.

A high volumetric muffler with low back pressure was also utilized to increase volumetric efficiency. This muffler was donated by Walker Dynamax.

One electronically preheated catalyst, specially formulated for use on a LPG vehicle was mounted in series with a non-preheated LPG catalyst. The catalysts utilized were manufactured by Emitec, Auburn Hills, Michigan, USA, and the coating of all precious metals was done by Engelhard, Iselin, New Jersey, USA.

The control for the heated catalytic converters lies in GMI's System Integration Module (SIM) which is discussed in a later section of this paper. The SIM controls the heating of the catalysts, and the vehicle's fuel purge system.

Figure 2: Catalytic Converer Efficiency [6]

GMI's 1997 Propane Vehicle Challenge(PVC) entry uses a liquid propane fuel injection system similar to last year's entry. While the number of different systems available are increasing, the injector and rail design we used (Fig. 3) best suited our design strategy of maintaining full engine control by the Powertrain Control Module(PCM). The injectors and rails, manufactured by BDE Ltd. of Lake Lillian, Minnesota, are mounted in the stock locations. The injectors are saturation type and therefore require no alteration of the signal from the PCM.

The fuel injectors, through the use of interchangeable washers, can easily have their dynamic flow rates adjusted as needed. Through the use of an oscilloscope, and the 0₂ sensor, the calibration best suited for GMI's system requirements was determined (Fig. 4).

This system, as with all liquid injection systems, must also be able to keep the propane in its liquid state. This is accomplished by circulating excess LPG throughout the injectors and fuel rails, and using the tank as a heat sink. It has been determined, though, that the amount of heat which we need to dissipate exceeds our tanks ability to do so. For this reason we have developed a system to cool LPG returning to the tank. Our strategy to dissipate the heat is discussed in the fuel delivery section.

To better understand this heat transfer phenomena an analysis of the fuel rail and injectors was performed.

Figure 3: Injector and Fuel Rail Profile [7].

Specifically, the fuel rail hold downs and the injector's lower housing were studied. The goal was to reduce the heat transfer to the LPG such that less energy would be returned to the fuel tank.

A finite element analysis of the fuel rail hold downs provided the team with temperature profiles for several different materials. By changing the hold down material from aluminum to nylon11 we were able to reduce the heat transfer by 85%.

During the last competition the GMI team used aluminum hold downs; after this analysis it was concluded that a lower thermal conductivity material should be used for these components. To remove the heat that is transferred to the injection system, the amount of fuel that is flushed through the system during the purge cycle is increased. This flow returning from the injectors to the tank was increased from 35 cc/min to 90 cc/min. This design change made it all the more prudent that a system be developed for managing this excess heat.

GMI's design goals included trying to improve the current controls in the vehicle to better suit propane injection. The goal was to significantly reduce the amount of time necessary to go closed-loop (when the air/fuel mixture is modified by the oxygen sensor output.)

This process was begun by obtaining two Fast Light Off Planar Oxygen Sensors, from Robert Bosch Corporation, Farmington Hills, Michigan. They are electrically heated, and use the stock wiring. The sensor incorporates two flat ceramic plates. This type Of O₂ sensor design requires only 5 to 10 seconds to warm up at room temperature, providing information to go closed-loop at an earlier point during operation.

In order to prevent severe cold start enrichment a thermistor was installed in series with the coolant temperature sensor (since propane vaporizes with air extremely well, even at subzero temperatures). This adjusts the temperature reading during cold weather startup so that the PCM senses a temperature reading which is higher than ambient. This thermistor does not affect the readings at high operating temperatures.

The SIM manages the purging of the injectors and fuel rails, the fuel delivery, and the preheat of the catalytic converter. it is a Motorola 66HC11 board, which has been programmed for optimum performance.

Due to the design of the injection system, liquid must be injected at all times, or the engine will fail to run. Using a temperature sensor on the fuel rail and a pressure sensor on the supply distribution block, the GMI-designed-and-built SIM determines the state of the fuel based on a thermodynamic table embedded in its software. This table is set approximately 10% lower than propane's actual thermodynamic properties in order to ensure that the fuel is in fact in the liquid state, given the error possible in the transducers.

If the propane is in the gaseous state, as determined by the GMI software, then the SIM will activate both of the fuel pumps, and purge the gas from the fuel rails until the readings show that the fuel is liquid. During this period the ignition is disabled, and an LED mounted within the dash makes the operator of the vehicle aware that the vehicle is purging. When the state of the fuel has changed, the additional pump is deactivated, the ignition is no longer disabled, and the voice emulator tells the operator it is no longer purging and may be started.

The initial pressure and temperature readings are only taken when necessary, as deemed by the Finite State Machine (FSM) logic which we have used to reduce the amount of time a driver may have to wait for the system to purge.

During normal operating conditions it is not necessary to take these readings and purge, as the excess LPG which is provided by our primary pump is enough to cool the injectors and fuel rails. This heated propane, which may be gaseous, is then returned to the tank.

In addition, the SIM controls the preheated catalyst. It uses the same FSM logic as used by purge system to determine when it is appropriate to begin preheating the catalyst, and for how long.

Due to LPG's high storage pressure, and unique characteristics it was necessary to totally redesign the current fuel delivery/storage system.

Through a joint effort (between Thiokol, Salt Lake City, Utah, Sleegers, London, Ontario, other competing universities, and ourselves) we have developed a comformable tank to store LPG. It is manufactured from aluminum extrusions which are currently being used for Chrysler's propane powered B van. This was done to decrease manufacturing costs, the tanks overall weight, and increase the amount of heat transfer versus a steel tank. By using B-van extrusions, this design also addresses the production needs for the near future.

A primary goal of our overall tank design was to maintain the stock position. Once again, this contributes to the production feasibility of the tank, and ease of installation.

In order to transfer the LPG from the tank to the engine we have incorporated a Walbro Engine Management, Cass City, Michigan, Liquid Propane Tank System (LPTS) This system consists of a circular aluminum casting which bolts to a similarly shaped flange on the tank (see Fig 5). The entire system consists of the following components:

• A level sensor which provides the signal for the stock fuel gage

• A single point reed switch which works in conjunction with a solenoid actuated valve to operate as an electronic stop fill valve

• A 2154 kPa pressure relief valve to protect against over pressurization

• A vapor drain port to allow for emptying of the tank for service

• A fusible plug to relieve pressure in the event of a fire

• One electric fuel pump for normal operation

• One electric fuel pump for use upon our SIMs demand

• One excess flow valve located on each fuel pump to cease the flow of fuel in the event of a line break

• One pressure relief valve to vent fuel, trapped between the fuel pumps and the supply solenoid, back to the tank

• One back flow check valve, located in the return port, to prevent fuel in the tank from entering the return line

From last year's competition we learned that the heat which we were collecting during normal operation, was not fully being dissipated before it reached the tank. By raising the temperature of the tank we had also raised the pressure to a point which opened our pressure relief valve. For this year's competition we have introduced a cooling system which will ensure that the propane which we are recirculating to the tank is at or near ambient temperature through the use of a Propane Cooling System (PCS.)

Figure 5: LPTS Diagram [8]

Inline on the return line we have introduced a propane to water heat exchanger manufactured by Edward's Engineering. From there the water is pumped to an air/water heat exchanger which has been mounted below the rear AC condenser on the front of the vehicle with virtually no modification.

In addition, flanged extrusions have been mounted at the base of the tank to evenly distribute the return fuel throughout the tank. This will allow us to better utilize the high thermal conductivity of the aluminum tank and reduce the amount of heat retained by the fuel as it enters the tank.

To enhance occupant safety many design changes have been incorporated. We have installed a three axis inertia switch, which shuts down our fuel delivery system in the event of a collision. We have incorporated a limit switch on the fuel door into our stop fill solenoid. This alleviates the potential improper refueling. A single point reed switch wired to the stop fill solenoid immediately discontinues fueling once the tank is 80% full. A 2154 kPa pressure relief valve has been installed in order to protect the system from becoming over pressurized. A fusible plug relieves pressure inside the tank in the event of a fire. An excess flow valve is located on each fuel pump. These valves detect when an inordinate amount of fuel is being supplied to the system, and immediately cut-off the flow of fuel. In the event that a line ruptures, these valves protect the occupant and the environment from dispersing fuel directly into the atmosphere. One pressure relief valve has been installed between the tank and the supply solenoid. This prevents the trapping of fuel between two points in the system. All on board safety systems have been retained with no modification. In addition, the total elimination of evaporative emissions controls is both a cost savings, and a safety improvement.

GMI's strategy throughout has been to maintain a system which will perform under any conditions. For hot start conditions the following modifications were made to insure consistent performance:

• Installation of temperature and pressure transducers
• to determine the state of the fuel

• Production and programming of GMI's SIM to control purge operations

• Software developed to minimize purge time

• Increased excess LPG flow through the PCS, injectors, and fuel rails

Addressing cold starting required only the use of the thermistor to modify coolant temperature readings.

The results for the cold start portion of the competition were exactly as we had anticipated. The cold-start was comparable to that of a normal gasoline system. However, the hot start competition revealed a minor flaw in our logic, as well as a minor electrical component failure. Both of these were remedied very quickly following this portion of the competition, with no further problems.

The system is designed for side by side production with today's gasoline powered vehicle at a rate of 10,000 vehicles per year. In order to maintain production feasibility we took the approach of modifying existing systems as minimally as possible without creating a multitude of systems which would be exclusive to LPG. In accordance with that goal we did the following:

• The tank is completely conformable in the same mounting position

• The injectors and fuel rails mount into the same locations with no modification to the mounts or PCM controls

• The SIM is a simple black box which operates in parallel with the current control systems

• The O₂ sensors will mount into the same locations

• The pistons are visibly different, yet require no modifications to the cylinder or rods

• All additional coatings were done to existing parts with little or no modification

The system is production ready today, and requires only minimal additional labor in production for installation of the coolant circuit, and the proper assembly of LPG fittings.

GMI's strategy in improving fuel economy centers around three key points: increased air mass flow, improved heat transfer characteristics, and an engine which takes advantage of the fuels high octane number in LPG. Accomplishing this goal required the following:

• Ceramic coating the intake

• A phenolic spacer to isolate the intake from engine heat

• Teflon coating the throttle body

• Profiling the throttle plate

• Port matching the intake system

• Increasing the compression ratio

• Reducing the amount of scavenging in the combustion chamber by increasing exhaust diameter

• Injecting LPG in its liquid state

These changes should have produced two results: an increase in fuel economy and better performance. The fuel economy competition yield no conclusive results due to difficulties with measuring equipment. The results for the 1/4 mile acceleration event revealed that this particular approach to the competition was successful for GMI, with a time of 17.935 seconds.

The approach to emission control included decreasing the amount of time it takes the engine to go closed loop, and increasing the total efficiency of our catalytic conversion system. We did this by:

• Ceramic coating of the exhaust runners

• Using a preheated catalytic converter
• Using a close coupled design for two catalytic converters
• Using 0₂ sensors which takes less time to warm up

These changes should allow our vehicle to surpass the emission requirements set forth by the Ultra Low Emission Vehicle (ULEV) standards.

The competion results for this vehicle were congruent with our expectations, yielding the results shown in Table 4. It should be noted, however, that by using a preheated catalyst GMI has neglected the long term implications of battery life in favor of improved emission characteristics.

Table 4: emission results

GMI's goal was to produce a system which would be production feasible at a rate of 10,000 vehicles per year, while maintaining an initial consumer cost penalty of less than $2,500 (Table 5). At a rate of 10,000 vehicles per year, the system which we have designed and manufactured should meet that goal. The most expensive components are the tank and injection components which are priced by volume. By using the same extrusions as those in the B-van tank we have tried to minimize any costs which may be incurred through a radically.

Table 5: Cost Penalty

different design. By using injectors compatible with the current PCM injector drivers, we have offset any cost increase associated with running a second production PCM.

This year's PVC entry for GMI is a product designed exclusively for the use of LPG. This design provides a means to work within the economical boundaries provided with such a large petroleum power base, while allowing for a vehicle which is truly optimized to operate as an alternatively fueled vehicle. It is through projects such as this that students, teachers, and industry can realize the full potential that all of our resources, from LPG to young engineers, have to offer the future of the automotive industry.

GMI's entry only won two of the twelve events in the competion. However, the entry did compete successfully over the entire competition, never placing lower than fourth, with exception to the Hot Start event. The results for the emission event are extremely promising, and could be pursued further with the proper funding and support. The team's low standing in the hot start event forced an improvement to the design of the vehicle, making the overall design more robust. This design still has many possible improvements, but it demonstrates some of the potentials this technology may bear for the automotive industry.

The GMI LPG team would first like to thank all of the organizers and sponsors of the PVC Challenge for arranging and coordinating such a large and educational event. We would also like to recognize all who contributed to GMI's successful completion of this project through their donations of time, materials, and money. Special thanks also goes to the competing teams who made the competition successful.

We gratefully acknowledge the contributions of the following companies for their material, financial and technical support: BDE Limited, Robert Bosch Corp., Bridgestone/Firestone, Emitec Corp., Engelhard Corp., Gillroy's Hardware, ITW Shakeproof, K & N Filters, Michigan Automotive Research Corporation, Magnacore, Motorola, Sleegers Engineering, Thermogas, Thiokol, Vetronix Corp., Walbro Engine Management, Edward's Engineering, George's Speedshop, Reliable, and Ross Racing Pistons.

Further thanks goes to several departments at GMI: The presidents office; for funding our project (and getting us to competition) and both the Mechanical and Electrical Engineering departments for providing us with work space and support.

Special thanks also goes to Drs.: Pinhas Barak, Behrouz Cheroudi, Colin Jordan, Daniel Sullivan and Etim Ubong, for their assistance on many technical issues.

The members of the GMI team were: Eric Anderson, Joseph Bouboulis, Colin Botha, Steven Bothe, John Celmins, Jason Dalton, Corey Keeslar, Timothy Kish, Denis Klug, Anne Orris, Todd Tischler, Joseph Titlow, Slaven Sljivar, Jason Trotter, and Henry Wandrie.