Proceedings of KGCOE-MD2003: Multi-Disciplinary Engineering Design Conference Page 5

Paper Number: MD2003-02008

Design of a Miniature Turbine for Power Generation on Micro Air Vehicles

Dan Holt / Arman Altincatal
Rob Latour / Carl Crawford / Srujan Behuria

Advised by Dr. Jeffery Kozak

Proceedings of KGCOE-MD2003: Multi-Disciplinary Engineering Design Conference Page 5

Abstract

This paper summarizes the progress completed by the Senior Design Team on a miniature turbine that is currently being developed at the Rochester Institute of Technology. The goal of this project is to design, fabricate, and test the casing and electrical components of a miniature turbine built to power small vehicles such as a micro air vehicle (MAV). MAV’s are vehicles with a maximum dimension of less then six inches and are used for surveillance and scouting. Due to the small size of MAV’s, weight is a key design parameter. The batteries alone can account for more than 50% of the weight of the entire vehicle. A miniature turbine driven by compressed gas and coupled to a DC generator has been proposed as a replacement for batteries on these and other small vehicles. The turbine would not only produce power for the electrical systems, but mechanical power for propulsion systems as well. This would eliminate the need for a motor. By decreasing the weight of power systems, MAV’s will be able to carry more instrumentation, fly longer, and be better able to complete their mission.

When fabricated, the turbine should spin at a minimum speed of 50,000 rpm and produce at least 5 watts of electrical power, but analysis shows that the turbine may reach speeds up to 100,000 rpm and produce up to 15 watts of power. Testing has begun to find the efficiency of the turbine as well as to compare the power to weight ratio to that of batteries.

Introduction

As Micro Air Vehicles (MAVs) and other small autonomous vehicles have gotten smaller, the proportion of the entire vehicles weight taken up by the power systems has gotten significantly larger. Batteries can take up to 50% of the weight of current MAV’s. A miniature turbine has been proposed as a solution to this problem. A number of schools are working on miniature (fabricated with conventional techniques) and micro (fabricated with semiconductor techniques) turbines. Miniature turbines have been fabricated on the 10 mm scale that produce 16 watts of power at around 100,000 rpm [1]. Micro turbines still face a number of challenges including bearings which will spin at 1 million rpm, combustion in the short allowable time, and production of enough power for MAV’s [2,3]. The goal of the work described in this paper is not only to build a power source for MAV’s but also to design a prototype that can be scaled down to the micro turbine size. Since the turbine under development uses compressed air as opposed to combustion to produce power and the blade geometry is less complex, the design is simpler then the designs being developed at other institutes. This paper outlines the design of a miniature turbine casing and electronics. Experimentation is ongoing to find the turbine characteristics and to compare these with similar commercially available dental turbines. A computational fluid dynamics (CFD) analysis is being developed to help future designers of miniature turbines.

Design

Several types of turbines were investigated for use as a miniature turbine for MAV power systems. An impulse, or Pelton wheel, type was chosen for several reasons. This type of turbine was initially based on the expected specific speed of the turbine. Specific speed is a nondimensional number used to compare different types of turbines to each other. Compressed air, nitrogen, or carbon dioxide was chosen to be the working fluid for the miniature turbine because it is inexpensive and inert. The gas will be inserted into the fuel tank in liquid form and then allowed to warm up to room temperature providing a high up stream pressure, up to a possible 46,000 psi. The pressure combined with a speed of 50,000 rpm and the small diameter gives a specific speed less than one. For this specific speed, impulse turbines are the most efficient [4]. Figure 1 shows the initial concept for the miniature turbine.

A second reason for choosing an impulse design was the possibility of scaling the turbine down to a micro turbine where semiconductor fabrication techniques must be used to build the turbine. These techniques severely restrict the design and make a radial or axial design prohibitive. Once the turbine type, working fluid, speed, and size of the turbine had been decided upon, a design could be done. The casing and electronics were designed to be as small as possible but produce the power necessary for MAV’s.

Casing Design

A casing was designed to accomplish several objectives. The air needs to flow into the casing and be directed at the impeller by one or more nozzles. The impeller shaft needed to be supported by two bearings that would be housed in the casing. Finally, the casing also had to be designed to minimize losses but also with fabrication difficulties in mind.

The final design of the casing uses three plates. The top plate is attached to the fitting for the air to enter the casing. The flow is split in the top plate to flow outward from the impeller shaft. Figure 2 shows the flow paths within the casing. Once the flow is outside the radius of the impeller, it is turned 90° and the leaves the top plate parallel to the shaft. The air then enters the middle plate, which supports one of the bearings. The air flows straight through the middle plate and into the bottom plate. There the air turns toward the impeller and flows through a nozzle that directs the flow toward the blade. The bottom plate supports the other bearing and contains the outlets for the turbine. This design was chosen largely due to its ease of fabrication since the flow paths can be milled into the aluminum plates. Since both flow paths are identical, the air exiting both nozzles should be identical. The use of one nozzle or multiple nozzles was investigated. More nozzles provide a larger torque on the turbine and balance the turbine but also greatly increase the complexity of the design. It was decided that two nozzles would be an adequate balance for the prototype miniature turbine.

The nozzles were designed assuming incompressible flow. The exit velocity of the nozzle is designed to be 100 m/s. This is just below the Mach 0.3 limit for incompressible flows, which simplified the design. With the knowledge of the length of tubing, the number of turns, and the upstream pressure the area ratio of the nozzle could be found. A CFD model, which is discussed later, was used to determine the nozzle placement, angle, and location of the exit ports. This completed the design of the casing.

Electronics

The electrical scope of the miniature turbine senior design project includes providing sufficient means to convert the mechanical power created by the primary mover, the turbine assembly, into electrical power. The long-term goal of this proof of concept is to utilize the design within the Micro Air Vehicle (MAV) program. The electrical design was carried out with this future goal in mind. The project needs assessment dictates that the miniature turbine shall produce at a minimum sufficient power for the electrical components of the current generation RIT MAV design. The power consumption of the current generation MAV, including the motor, is approximately 3 Watts.

To facilitate the conversion from mechanical to electrical power a direct current (DC) motor will be employed, specifically a permanent magnet dc motor (PMDC). No physical differences between a DC motor and a DC generator exist, the only distinction being the direction of power flow. In a DC motor electric power is converted to mechanical power. Conversely, in a DC generator, mechanical power is converted to electrical power.

Anther benefit of the DC motors is that they are two-way machines allowing the motor specifications supplied by the manufacturer to be directly applied to the same motor acting as a generator. This allows the turbine speed and torque to be measured as will be discussed in the following section.

The Faulhaber 1628…B series brushless DC motor will be utilized in this experiment. The motor is rated at 10 W, has a maximum speed of 65,000 rpm, is 28 millimeters in length and has a diameter of 16 millimeters.

Experimentation

Experiments are being conducted to develop the turbine characteristics. Torque, power, and efficiency will be found versus mass flow rate. These curves will also be found on two dental turbines. Dental turbines are impulse turbines designed for very high speeds, 300,000-500,000 rpm, but to produce very low torques so they cut into teeth slowly. This provides a comparison for the turbine described in this paper to a similar commercially available turbine.

In order to find the torque and power an electric motor, or generator, has been attached to the turbine. The output power from the motor is recorded at various flow rates. From the generators output voltage, the rpm of the motor and therefore of the turbine, can be determined. The torque can be found by dividing the power by the rotational speed and assuming a small loss in the coupling. The efficiency of the turbine will also be determined. This will be done by setting up a pressure sensor and two temperature sensors (thermocouples). Figure 3 shows the experimental setup. The air starts in a compressed air tank where the flow is controlled. After leaving the tank, the air enters a plenum where it settles and the temperature and pressure can be measured.

Once the air has left the plenum it travels through a flow sensor, which measures the volumetric flow rate. Using the temperature (T), pressure (P), and volumetric flow rate (Q), the mass flow rate (m dot) can be calculated using Equation 7:

(7)

The mass flow rate will be used to calculate the mechanical power produced by the turbine. Once the air exits the flow sensor, the air enters the turbine where it spins the impeller and then exits the casing. The exit temperature of the air is also measured.

Figure 3

The efficiency of the turbine is equal to the actual work (W) divided by the isentropic, or ideal, work. The actual work is equal to the torque times the rotational velocity, both of which will be measured using the generator. The ideal work can be found using one of two different methods depending on the flow conditions. If the flow entering the turbine is compressible, defined by the Mach number being greater then 0.3, then the ideal work can be found using the change in temperature and pressure across the turbine and (k) the ratio of specific heats as shown in Equation 8.

(8)

If the flow entering the turbine is incompressible, a Mach number less then 0.3, then the ideal work can be found using the pressure and velocity (V) in Equation 9.

(9)

For both of these equations, the downstream pressure is assumed equal to atmospheric pressure. These measurements will provide the data to develop a map of the turbine.

Initial testing has been completed. The max power produced by the turbine to date is 3.2 Watts at ____kPa (140 psi) upstream pressure. This was produced at slower speeds, 10,000 to 15,000 rpm, due to limitations of the generator. The turbine has been spun up to 50,000 rpm with no load. The generator dicussed above is currently being set up and should allow operation at higher speeds. Since predictions show that the turbine will be more efficient at higher speeds, the turbine should produce more power.

CFD Analysis

A CFD analysis is being developed to help future designers of miniature turbines. The code attempts to model the flows inside the turbine casing. This model will then be compared to the experimental data to try to improve the accuracy of the model. Future designers can then modify the geometry of the casing and the turbine to qualitatively estimate the affects the changes will have on the turbine.

Figure 4 – Geometry and Grid

Figure 5 – Velocity Vectors (m/s)

A number of steps have been taken to build the current 3-D CFD model. Initially a simple 2-D model was created to learn how to model a rotating geometry. As the simple model evolved it was used to help design the casing by modifying the geometry of the nozzle placement and angle at which the air impacts the blade. As the model developed further it also provided an estimate of the torque produced by the turbine. Based on a required torque to spin the generator, the inlet velocity, upstream pressure, and flow rate could be estimated. The model also provided an estimate of the velocities and pressures inside the casing. Figures 4, 5, and 6 show the model geometry, velocity vectors, and pressure distribution.

Figure 6 – Dynamic Pressures (Pa)

A 3-D model was then built based on the 2-D geometry. A 3-D geometry required significantly more nodes (points where equations are solved to determine flow properties) to be able to solve the flow field efficiently. Boundary adaptation was used to increase the number of nodes in the critical areas such as blade surfaces and the blade tips. The total number of computational cells used in the model was approximately 770,000 tetrahedral elements. The model was at the limits of the computers memory capabilities (> 800,000 elements). Figure 6 shows the mesh on a center plane near the surface of the top blade.

Figure 6 – Mesh Details

The fluid regions between the blades were modeled as volumes to apply the Moving Reference Frame (MFR) to account for the rotational effects [5]. The surrounding surfaces such as the blades and the turbine were defined as rotational surfaces relative to the volume regions. Figure 7 displays the 3-D geometry of the casing and the turbine obtained from the post-processor Fluent 6.0.