UAB - ECTC 2014 PROCEEDINGS - Section 7 Page

SECTION 7

DESIGN & MANUFACTURING

UAB School of Engineering - ECTC 2014 Proceedings - Vol. 13 179


Proceedings of the Fourteenth Annual Early Career Technical Conference The University of Alabama, Birmingham ECTC 2014 November 1 – 2, 2014 - Birmingham, Alabama USA

PRELIMINARY COMMERCIAL DESIGN FOR TRANSPORTING LOW TEMPERATURE PARTICLES FOR USE IN PARTICLE SOLAR RECEIVERS USING

LIFTING SKIPS

Kenzo Repole, Sheldon Jeter Georgia Institute of Technology

Atlanta, GA, USA

ABSTRACT A particle heating receiver (PHR) system is a form of

concentrating solar power (CSP) system that has the ability to incorporate thermal energy storage (TES).

The particles can then be directed towards a heat exchanger to a power generation system or to storage for later use when the utility load demand is greater or as a means to compensate for the solar variation experienced during the day.

As the need for greater capacity CSP and especially PHR systems grows, so will the need to find an effective delivery system to dispatch the working particles to the PHR at faster rates and in larger amounts.

We have shown that out of the many different systems that can be used to transport the particles up to the PHR, a suitable commercial solution would be a Kimberly skip (KS) based particle lift. This design would be low cost in comparison to the CSP power capacity, it would have a high overall efficiency and service life and low maintenance cost.

INTRODUCTION A particle heating receiver (PHR) system is a form of

concentrating solar power (CSP) system that has the ability to incorporate thermal energy storage (TES).

In a PHR system, solar energy is irradiated from the field of heliostats onto a falling curtain of particles at the aperture location of a power tower.

The particles can then be directed towards a heat exchanger to a power generation system or to storage for later use when the utility load demand is greater or as a means to compensate for the solar variation experienced during the day.

The advantage that PHR has over other forms of CSP is the ability use low cost materials to store sensible and latent heat over a longer period of time and where the degradation of the working material is greatly reduced over the life of the PHR helping to reduce the levelized cost of energy (LCOE)[1].

As the need for greater capacity CSP and especially PHR systems grow, so will the need to find an effective delivery system to dispatch the working particles to PHR at faster rates and in larger amounts.

Larger capacity PHR systems with TES means the power tower will increase in width and especially in height in order to store the high temperature particles used during power generation or as storage for demand based power generation.

Currently, there are many different methods of carrying fine particles up to elevations of the PHR. However, to increase the capacity of the power plant and its efficiency, the particles, entering the PHR, need to be at higher temperatures ranging from 300°C (572°F) up to 600°C (1112°F).

At this temperature more conventional methods of delivery of large amounts of working particles are not viable, since the operating environment is outside the operating range for delivery mechanism.

PARTICLE TRANSPORTATION Currently, three options are available to meet this

challenge. They are bucket elevators, Olds Elevators, and particle skips similar to those used in the mining industry.

The bucket elevators have the ability operate at high temperatures, greater than 200°C (392°F) [7]. However, it would require the shaft used in the particle transport to be kept at the same temperature as the particles in order to maintain a viable thermal efficiency for the power system. Moreover, this system would experience a high spillage rate during operation.

An Olds elevator (OLDS) system has the ability to deliver the working particles in a continuous flow. However, as the height of the tower increases due to increased capacity demands, the cost of the OLDS system increases significantly due to the nature of its design.

A suitable design option that can address the current and future needs of larger capacity PHR and that maintains high thermal efficiency and low exergy degradation would be a particle lift.


PARTICLE SKIP LIFT Particle lifts similar to those used in the mining industry

come in different forms. The main forms currently used in the mining industry are Bottom Dump skips, Front Dump skips, Kimberly Skips (KS) and Arc Gate skips.

The main tradeoffs between the different forms of skips involve the extra height required during operation versus the ease of operation at high temperature versus spillage during use.

Bottom dump skips are charged from the top and discharged by a trap door at the bottom of the skip. They do not require a large amount of extra height for operation in comparison to the other types of skips. This skip design is light weight and rugged, but due to the fine size of the particles used in the PHR, spillage will be large during the transport and discharge of the particles.

Front dump skips, are charged from the top and discharged by rotating the entire skip to an incline where the particles are discharged through a gate at the lower end of the front side of the skip. They can carry large volumes of particles and put the least amount of stress on the head frame. However, the spillage rate is still high in comparison to the other types of skips.

Arc Gate skips are considered the safest and most rugged in industry. They are charged from the side and discharged through an arc gate on the side. As with the bottom dump skip, the arc skip will experience large amounts of spillage during charge and discharge. In addition to this, the arc gate has many parts, which means an increased risk of failure under high

temperature extreme environments that would be experienced in transporting the working particles in the PHR.

Kimberly skips are charged and discharged from the top side of the skip. The particles are loaded into the skip from the top. The skip then travels in this configuration only it reaches its discharge location. As it reaches the discharge location, a set of scroll wheels on the skip engages scroll guides on the shaft walls. These wheels guide the skip and allow the skip to rotate to about 120° from its vertical position. This action discharges the particles from the skip. A KS can carry large loads, but Front dump skips can carry about four times the amount in comparison for similar designs. The KS has the lowest initial cost, the lowest maintenance cost and highest service life in comparison to the other skip types [2]. KSs have the lowest amount of spillage occurrence during use. KSs, however, require the largest headroom and width clearance of any skip design. They also exert the largest amount of stress on the head frame [3, 4].

COMMERCIAL SCALED PARTICLE LIFT Our goal was to develop a suitable commercial particle lift

solution. The conceptual design and engineering of a commercial particle lift system, including the completion of a small-scale model of the proposed design, was completed. Conceptual design drawings and energy efficiency and heat loss modeling have also been completed. Detailed efficiency modeling based on reliable published component efficiencies resulted in an energy efficiency of 80%, which exceeds the

Figure 1. Insulated Kimberly skip charging (left), travel position (center) and discharging (right).


75% energy efficiency metric. With this efficiency, the parasitic power should be less than 1% of the rated output. The selected skip design, was that of the Kimberly skip type, seen in Figure 1 with its charging, travel and discharge configurations.

This skip is both filled and discharged from the top and has no complicated and leak-prone bottom hatch. This arrangement facilitates a design that is very simple structurally and mechanically. The single top hatch, which is opened and closed by motion of the skip thereby eliminating any mechanical or

hydraulic actuators, is critical to this simplicity. Importantly, this design appears to be almost leak proof and should easily achieve much less than 0.1% temporary spillage of particulate during filling and discharge.

All such spillage will be accumulated in a sump built into the lift shaft, which can be emptied as necessary; therefore, there will essentially zero net loss of particulate from the system. Minimal heat leak is also an objective, and simplicity of the proposed skip design makes it easy and inexpensive to install adequate internal insulation to keep the heat leak from the skip well under 0.1% of the rated capacity of the system.

Our design also envisions an elevator shaft allowed to stay at 200°C (392°F), which further minimizes incidental heat leaks. Altogether the proposed design ensures a minimal heat leak that will have negligible effect on the overall system efficiency.

The first set of drawings and specifications have been completed, and we have already consulted with one company familiar with steel fabrication and industrial lift manufacture. This company has commented that our design will be easy to manufacture. After incorporating some minor modifications based on this review, we will be consulting with a skip-hoist component supplier. With helpful input from these initial reviews from smaller companies, we will be contacting one of the major manufacturers.

Initially, many lift options were considered. In first phase of the project, we identified the counterbalanced skip hoist as the most promising on efficiency, cost, and reliability. Two generic types were considered in the second phase of the project: (1) the Front Dump Skip and (2) the Kimberly Skip. The Front Dump Skip is evidently favored in traditional mining since its layout is compatible with a relatively small cross section and longer length. The smaller cross section is highly desirable in mining where the vertical shaft can be hundreds to thousands of meters deep. In contrast, the simplicity of the Kimberly skip promotes a low initial cost, low maintenance cost and high service life.

All features are important in CSP applications. In particular, the simple design (basically a bucket with a hinged lid and a lifting bail) allows effective thermal insulation with mere layers of continuous suitable rigid insulation such as firebrick inside the skip and the lid with no complicated bottom hatch to insulate and no mechanism components (such as links and latches) likely to act as thermal short circuits. The bottom hatch is also likely to leak during lifting, which is not an issue when handling typical raw materials but important when hoisting the TES medium. Our experience with the two small-scale models was convincing with regard to these issues.

As shown Figure 2, the Kimberly skip is easy to integrate into the CSP system. Note that the lift shaft will be kept at elevated temperature around 200°C (392°F) to minimize heat losses, but the electrical and mechanical equipment (other than the lift drum) will be kept at near ambient temperature for efficiency and economy.

No. Name 1 Lift Machine Room 2 Lift Discharge Chute 3 Particle receiver

4 High Temperature TES Bin

5 PWFHX

6 Low Temperature TES Bin

7 Lift Charge Chute 8 Lift Shaft 9 Top hopper

Figure 2. Lift integrated into TES Tower.


Typically, stainless steel such as SS316 wire rope is selected for durability, corrosion resistance, and excellent high temperature strength. Cost estimates based on a highly regarded source of generic unit and subsystem costs are summarized in the following table:

Table 1. Current Cost Analysis for Particle Lift

Estimated Cost of 2 Skips without Hoist System $198,533 Estimated Cost of Hoist System $294,600 OLDS Elevator Particulate Recovery system $30,000 Total Estimated Cost per System $523,133 Total Estimated Particle Lift cost per MWth $8,719

Accordingly, the lift system is expected to cost around

$8,719 per MWth, which agrees with previous independent cost estimates using technology-specific cost engineering research results, which was around $9,614 per MWth [5].

Details of the design and modeling will be the subject of technical papers, two of which are now being prepared. This analysis was conducted assuming a module of 60 MWth. A system size up to 460 MWth would be accommodated by some combination of larger skips and multiple pairs of skips.

Typically, costs per unit are improved at larger size and overall efficiency is only negligibly changed with larger system size and longer lift. For a 460 MWth system, with a vertical height of 138 m (453 ft), some particular critical stress values are as follows: calculated critical stress in skip metal would be 6.48 MPa; design stress in wire lift rope would be 112 MPa.

The rope size of 0.076 m (3 inch) based on the above calculated stress and on vendor tensile strength of SS316 using the factor of safety of 5 as required by OSHA[6] for wire rope and many other critical applications.

The energy efficiency modeling is based on lift and recovery efficiencies and the ratio of overall tare to payload in Table 2. The tare fraction is important since the potential energy of the skip and rope cannot be 100% recovered. Also, the lift efficiency of 85% is from several published standards and models and has been confirmed by component modeling shown in Table 3. The expected energy efficiency is 80%, which is higher than the target of 75%.

Table 2. Estimates of overall efficiency for particle lift design.

Data Efficiency Lift Efficiency 0.85

Recovery Efficiency 0.93 Ratio: Tare/PL 0.2549

Overall Efficiency 0.8064 Fraction Parasitic 0.0086

Table 3. Estimates of lift component efficiency for particle lift design.

Component Efficiency

VF Drive 0.96 Electric Motor 0.95

Gearing, 2 Stage 0.98 x 0.99 = 0.97 Rope/Drum Efficiency 0.98

Overall Product 0.86 to 0.87

CONCLUSIONS In conclusion, as the need for better performing and larger

capacity CSP increase, the need for lower cost and higher capacity TES will follow suit. This will be especially true for solid particle based TES.

We have shown that out of the many different systems that can be used to transport the particles up to the PHR, a suitable commercial solution would be a KS based particle lift. This design would low cost in comparison to the CSP power capacity, have a high overall efficiency and service life and low maintenance cost.

ACKNOWLEDGMENT Financial support of the US Department of Energy through the SunShot research program is recognized and appreciated.

REFERENCES [1] Ma Z, Glatzmaier G, Mehos M, Fludized Bed Technology for Concentrating Solar Power with Thermal Energy Storage. Journal of Solar Engineering 2014; 136 [2] Wabi Iron & Steel Corp. [3] De la Vergene, J., “Hard Rock Miner’s Handbook”, Edition 3, 2003, McIntosh Engineering, Ontario, Canada. [4] “SME Mining Engineering Handbook”, Third Edition, 2011, Society For Mining, Metallurgy, And Exploration Inc. [5] Internal Study. [6] OSHA “Safety And Health Regulations For Construction” 29 CFR 1926.251(2010) [7] Rexnord Corporation



PROPAGATOR 2: A PLANING AUTONOMOUS SURFACE VEHICLE WITH AZIMUTH RIM-DRIVEN THRUSTERS

Daniel Frank University of Florida

Gainesville, Florida, U.S.A.

Andrew Gray University of Florida


Dr. Eric Schwartz University of Florida


ABSTRACT This paper outlines the research and testing that went into

the mechanical design and implementation of PropaGator 2. The goal of this project was to develop a small-scale, high-speed, fully autonomous surface vehicle (ASV). For the purposes of this paper, small-scale is defined as displacing less than 100 𝑙𝑏𝑓 of water, while high-speed is defined as greater than 10 𝑘𝑡𝑠. In particular, this paper focuses on the design of the ASV’s hull, azimuth steering system, Kort nozzle, and rim-driven hubless propellers. There are many applications for such a platform including, search and rescue, reconnaissance, minefield investigation, and small payload transportation between surface vessels.

1. INTRODUCTION Autonomous vehicles have gained popularity in recent

years due to the advancements in technology. The processing capabilities of computers have increased with a concurrent decrease in cost. Sensors that were once uncommon have become more common due to the growth of the hobby robotics market. The autonomous vehicle population at the University of Florida (UF) reflects this growth.

At the UF Machine Intelligence Lab (MIL) students from electrical, mechanical, and computer engineering departments design autonomous vehicles for land, sea, and air. Recently, the lab has placed an emphasis on naval vessels. MIL’s first ASV, built in 2012-2013, was PropaGator 1 [1]. PropaGator 1 was designed with the intention of adding a surface vehicle to a heterogeneous robotic swarm [2]. The required capabilities of the ASV were to autonomously navigate an obstacle course in the water and to maneuver in any direction on the water’s surface. The ASV also was limited by weight and volume constraints of 120 𝑙𝑏𝑓 and 72 𝑖𝑛 x 36 𝑖𝑛 x 36 𝑖𝑛 respectively. These constraints were imposed to allow the boat to be eligible to compete in the 2014 AUVSI Foundation’s RoboBoat Competition. Due to the strict time constraint placed on the development of PropaGator 1, priority was placed on the rapid design and construction of the ASV.

It was necessary to further reduce the development time to allow sufficient time for testing. Like all displacement vehicles, PropaGator 1’s theoretical maximum velocity was determined by its hull speed. At hull speed, an ASV becomes trapped behind its own bow wave and is no longer able to increase its velocity. Hull speed can be estimated using the following formula [3], 𝑣ℎ𝑢𝑙𝑙 ≈ 1.34�𝐿𝑤𝑙 , (1) where 𝑣ℎ𝑢𝑙𝑙 is the hull speed measured in 𝑘𝑡𝑠 and 𝐿𝑤𝑙 is the length of the waterline measured in 𝑓𝑡. Using this formula, the maximum top speed of PropaGator 1 was predicted to be 3.14 𝑘𝑡𝑠.

While PropaGator 1 was a fully functional ASV, the vessel’s speed and weight limited its ability to complete long range missions. The combination of the drag caused by the hull and weak trolling motors prevented the ASV from navigating the obstacle course in a more timely fashion. Due to the design of the ASV, PropaGator 1 could not be easily modified for improvement. In order to achieve the desired performance, a new platform had to be developed.

PropaGator 2, as shown in Figure 1, was conceived to far exceed the performance of PropaGator 1. Without sacrificing maneuverability, the new ASV was designed to displace less than 100 𝑙𝑏𝑓 of water, travel at 10 𝑘𝑡𝑠, and still meet the size constraints of its predecessor.

Figure 1. CAD render of PropaGator 2


2. HYDROFOIL RESEARCH As shown in Equation 1, the maximum speed for

displacement vessels is dictated by a single geometric parameter, the length of the water line. Since a constraint of a maximum length of 72 𝑖𝑛 was imposed, the decision was made to design a displacement hull that featured hydrofoils. Like an aircraft wing, hydrofoils produce lift as the vehicle moves through the water. By partially raising the ASV out of the water, the ASV is allowed to ride over its bow wave and reach velocities greater than those determined by the hull speed. The hydrofoils were designed using an Eppler 817 hydrofoil cross section [4]. An image of the foil shape can be seen in Fig 2.

This cross section was chosen for its minimal disposition

towards cavitation and its ease of manufacturing. The foil was cut using a 4-axis CNC mill and tested for its lift and drag characteristics in a wind tunnel shown in Figure 3.

The goal of the test was to determine the angle of attack

that produced the greatest lift-to-drag ratio as well as the maximum lift force that the hydrofoil could generate while moving in the water at 10 𝑘𝑡𝑠. The air in the wind tunnel had a Mach number of approximately 0.2, therefore no significant effects due to compressibility were present. Water, under the ASV’s normal operating conditions, may also be assumed to be

incompressible. Since both fluids were incompressible and the Reynolds number of the ASV’s top speed was matched with the wind tunnel’s Reynolds number, the data collected in the test estimated that one hydrofoil, with an angle of attack of 5°, could produce 25.62 𝑙𝑏𝑓 of lift when the ASV is traveling at 10 𝑘𝑡𝑠 with a lift-to-drag ratio of 6.55. This would allow a 100 𝑙𝑏𝑓 ASV to completely lift out of the water when moving with a velocity of 10 𝑘𝑡𝑠.

After the wind tunnel results were analyzed, a wooden prototype hull with hydrofoils was constructed as shown in Figure 4. Two fixed 80 𝑙𝑏𝑓 thrust trolling motors were installed onto the prototype and a number of speed tests were conducted in order to determine its maximum velocity. The prototype was able to achieve a speed of 6 𝑘𝑡𝑠, nearly twice as fast as it would have been able to move if it didn’t have hydrofoils. However, it was not able to generate enough lift to completely rise out of the water. Lift can be calculated with the following equation [5],

𝐹𝑙 = 12𝜌𝑣2𝐴𝐶𝑙 , (2)

where 𝐹𝑙 is the lift force in 𝑙𝑏𝑓, 𝜌 is the density of water in 𝑙𝑏𝑚 𝑓𝑡3⁄ , 𝑣 is the velocity of the ASV in 𝑓𝑡 𝑠⁄ , 𝐴 is the reference area of the hydrofoil in 𝑓𝑡2, and 𝐶𝑙 is the dimensionless coefficient of lift.

The reason why the prototype was unable to completely lift

out of the water was due to drag. In order to quantify the amount of drag the prototype generated, it was towed fifteen feet off the port side of a larger boat. A load cell was used to measure the drag force acting on the prototype at a variety of velocities. Additionally, flow simulations were conducted in SolidWorks as another way to estimate the drag force. The results of both the empirical testing as well as the SolidWorks simulations can be seen in Figure 5. In general, the SolidWorks simulations coincided well with the measured values of drag

Figure 4. Underside of the wooden prototype hull featuring hydrofoils

Figure 3. Hydrofoil mounted in the wind tunnel

Figure 2. Eppler 817 cross section used to design the hydrofoils tested on the prototype


obtained in the drag test. Part of the reason that the SolidWorks values are higher than the empirical data is that SolidWorks did not factor in the reduced drag on the hull as the prototype was lifted out of the water. Even so, SolidWorks provided conservative estimates on drag and lift that helped in the design of our next and final hull.

While the hydrofoils did help lift the prototype out of the

water, causing a reduction of drag due to less wetted surface area, the largest contributors of drag came from the motors themselves, a fact that will be explained below. Drag may be calculated with the following equation,

𝐹𝑑 = 12𝜌𝑣2𝐴𝐶𝑑 , (3)

where 𝐹𝑑 is the force due the drag in 𝑙𝑏𝑓 , 𝐶𝑑 is the dimensionless coefficient of drag, and 𝜌, 𝑣, and 𝐴 have the same meanings and units as Equation 2. In this analysis, 𝐶𝑑 is assumed to be a function of the Reynolds number. Since drag is a function of 𝑣2, as the prototype increased its velocity, the drag caused by the motors increased with the velocity by a power of two. The decrease in drag caused by the lifting of the prototype was not able to offset the additional drag caused by the motors, thus the overall drag of the prototype was always increasing with velocity. As the velocity was increased, at a certain point, the available thrust produced by the motors became less than the total drag. Therefore the prototype was no longer able to accelerate and gain velocity.

3. HULL DESIGN After proving that the prototype would never be able to

achieve 10 𝑘𝑡𝑠, a new hull was designed. The upper half of the hull was modeled after the M80 Stiletto shown in Figure 6. The flat planes that make up the majority of the features on the top half of the ASV are aesthetically pleasing while also being easy to manufacture. The research performed on the prototype with hydrofoils helped motivate many of the design choices used on the submerged portions of the hull. As shown in Equation 1, the length of the waterline constrains the maximum speed a displacement style hull can achieve. However, the maximum velocity of a planing hull does not depend on its length. The

maximum speed of a planning vessel can be predicted using Crouch’s formula,

𝑣𝑝 = 𝐶

� 𝐷𝑆𝐻𝑃

, (4)

where 𝑣𝑝 is the maximum planing velocity of the boat measured in 𝑘𝑡𝑠, 𝐶 is a constant based on the hull form of the vessel, 𝐷 is the amount of water the vessel displaces in 𝑙𝑏𝑓, and 𝑆𝐻𝑃 is the shaft horsepower at the propeller. The displacement term is another justification on keeping the weight of the ASV to a minimum. To achieve this low weight, plugs for the ASV’s hull were cut out of EPS foam using a CNC mill. The plugs were then conditioned so that a fiberglass mold could be created. Once the molds were finished, the ASV was constructed out of fiberglass. The use of fiberglass gave the ASV structural strength while still being light weight.

A catamaran design was chosen over a monohull for

several reasons. First, catamarans offer more stability in the roll direction than a traditional monohull. Second, catamarans are also able to displace the same amount of water with a shallower draft compared to a comparable monohull, allowing an ASV to navigate shallow waters safely. Finally, catamarans also typically have the advantage of providing less resistance in the water than a comparably sized monohull [6]. The shallow deadrise helps provide a smooth ride to reduce noise for the on-board sensors while still providing a large planning surface to generate lift [7]. Hard chines were added to the side of the hull to give the ASV better forward tracking. The upper style chines also help redirect the water from riding up the side of the ASV which results in a reduction of drag. The V-shape front of the pontoons is designed to help break waves in choppy waters. Once the water is separated, the pontoons have a constant cross section until the stern of the ASV ends with a hard transom. The hard transom allows the water to separate from the ASV at a known point and helps to prevent issues with flow separation.

4. AZIMUTH STEERING SYSTEM PropaGator 1 featured four trolling motors that were

rotated at set angles of 30° from forward, allowing the ASV

Figure 6. U.S. Navy’s M80 Stiletto

Figure 5. Drag test and simulation results


three degrees of freedom; translation in two directions and heading. While this configuration can achieve motion in any direction, the ASV loses efficiency since all motions are a result of four thrust vectors with components in opposition. Since the ASV primarily moves forward, it makes sense to design the forward direction to be the most efficient direction for the ASV. By utilizing two azimuth thrusters, PropaGator 2 is able to achieve the same mobility as PropaGator 1’s four trolling motors [8]. An azimuth thruster is simply a marine propeller than can be rotated about the vertical axis to assist in steering. By going from four motors to just two steerable motors, the drag caused by the motors was reduced. Additionally, by using steerable motors, PropaGator 2 is more efficient than its predecessor, especially when moving forward and backwards.

5. PROPULSION SYSTEM The drag test experiments verified that the large trolling

motors were the largest contributors towards drag at high speeds. As a result, it was decided to design and manufacture a new propulsion system with no submerged motors. One of two identical thrusters for PropaGator 2 is shown in Figure 7. The goal was to minimize drag by shrinking the water footprint of the propulsion system. Rather than having the motors in the water like the motor pods developed by Hsieh et al. [9] or standard trolling motors, the motors are located inside of the hull. The power is transmitted from the motor to the propeller through a timing belt that is attached to the output shaft of the motor and the rim of the propeller. This rim-driven transmission eliminated any drag caused by the hub of a propeller.

5.1 KORT NOZZLE For safety reasons, it was decided that all propellers should

be shrouded. Unfortunately, adding a shroud leads to additional drag. By designing a propeller that is enclosed by a duct, also known as a Kort nozzle, it was possible to create a duct that helps generate high thrust at low speeds [10] while still producing minimal drag when moving through the water. Ducted-propellers can also help increase fuel efficiency by allowing the ASV to run the propellers at a lower RPM while still maintaining the same velocity as an equivalent ASV with non-ducted propellers. Furthermore, the ducts prevent prop walk, a phenomena that causes the stern of the boat to strafe in the direction of the propeller’s rotation. Typically twin-screw vessels, boats with a propeller on the starboard and port side, will have two different shaped propellers; one that rotates clockwise to achieve forward motion and one that does so by spinning counter-clockwise. The counter-rotating propellers balance out the forces that cause prop walk. However, by using the ducted-propellers, only one type of propeller needs to be manufactured, which allows for easier replacement if one breaks. 5.2 PROPELLER DESIGN

The motivation behind the development of the propulsion system was to reduce drag as much as possible. During simulations and testing, it was determined that the hub in a ducted propeller serves two purposes: to provide strength to the blade roots and to provide an attachment point for the motor. The hub does not produce any thrust, but it does produce drag. By migrating to a rim-driven design, the hub’s only remaining purpose was to provide strength to the blades. Unfortunately, the hub only offers strength to the root of the propeller, which in general, experiences loads that are much lower than those experienced by the tips [11]. By inverting the propeller into a hubless configuration as shown in Figure 8, the moment arm acting on the propeller tips is reduced, causing an overall reduction of bending moment on the rim-driven propeller compared to an equivalent standard hub-driven one. By eliminating the hub, the drag that it produces is eliminated. The hubless propeller also has the advantage of being resistant to propeller fouling due to debris [12].

The propeller blades themselves were designed using Crouch’s Method [13]. The maximum planing speed of the ASV was calculated using Equation 4. The required pitch to obtain that speed can then be found using,

𝑝 = 1215.22∙𝑣𝑝.9∙𝑅𝑃𝑀

, (5) where 𝑝 is the pitch of the propeller measured in 𝑖𝑛, 𝑅𝑃𝑀 is the maximum rotations per minute the motor is capable of spinning, and 𝑣𝑝 is calculated in Equation 4. This equation assumes that the operating speed of the boat occurs when the motors are outputting 90% of their total power and speed. The pitch is the theoretical distance that the propeller should move forward with one rotation, like a screw turning through wood. In practice, the propeller never moves as far forward as the Figure 7. Assembly of PropaGator 2’s propulsion system


pitch would suggest since the water slips around the propeller as it is spinning. The difference between how far the propeller should move in the water and how far that it actually moves is known as apparent slip. In order to accommodate for the presence of slip, the pitch must be made larger by multiplying it by an empirically determined correction factor.

After the pitch had been determined, the diameter of the

propeller was calculated. Ideally, since the goal is to reduce the drag caused by the propeller assembly, it would be best to make the propeller as small as possible. While a very small propeller could provide adequate thrust at high speeds, it may not be able to provide enough thrust at lower speeds to reach the planing state or to effectively maneuver. To determine the minimum diameter the following equation was used,

𝐷𝑚𝑖𝑛 = 4.07 ∙ �𝐵𝑊𝐿 ∙ 𝐻𝑑, (6) where 𝐷𝑚𝑖𝑛 is the minimum required diameter to provide useful thrust at all speeds measured in 𝑖𝑛, 𝐵𝑊𝐿 is the beam on the waterline measured in 𝑓𝑡, and 𝐻𝑑 is the draft of the hull measured in 𝑓𝑡. Since PropaGator 2 is designed to be a twin-screw vessel, 𝐷𝑚𝑖𝑛 must also be made smaller by being multiplied by an empirically determined correction factor.

Typically propellers with two blades are more efficient than propellers with more blades. However, in order to get the blade area required for effective thrust, they require a large diameter. Since the goal was to minimize the diameter of the propeller as much as possible, a three-bladed propeller was designed. The extra blade allows for additional blade area without increasing the propeller’s overall diameter. Other common propeller parameters are rake and skew, but after researching how they affect the performance of the propeller, it

was decided that their inclusion in the propeller design did not offer any advantage to the application of PropaGator 2.

Once all of the propeller parameters were determined, shown in Table I, a CAD model of the propeller was generated and then machined on a 4-axis CNC mill. In order to provide resistance to moisture, long-term fatigue endurance, and high strength and rigidity, the propellers were cut out of blocks of Delrin.

Parameter Value Number of blades 3

Diameter 3.5 𝑖𝑛 Pitch 7 𝑖𝑛 Rake 0° Skew 0°

Cross section shape NACA 66-006 Developed blade area 1.2 𝑖𝑛2

6. RESULTS The maximum velocity of PropaGator 2 is 11 𝑘𝑡𝑠 and was

recorded by using the velocity log from a Yuan 10 Skytraq S1315F-RAW GPS. The final displacement for the ASV was 95 𝑙𝑏𝑓. It can fit within a box with the dimensions of 72 𝑖𝑛 x 35 𝑖𝑛 x 21 𝑖𝑛 and has a maximum payload of 200 𝑙𝑏𝑓. Since the hull’s mold has already been fabricated, to build an exact replica of PropaGator 2 would cost less than $15,000.

In July of 2014, PropaGator 2 competed against other ASVs from around the world at AUVSI Foundation’s 7th Annual RoboBoat Competition. In addition to winning the Innovation Award for Hull Form and Propulsion Design, it outperformed all of the other ASVs in the propulsion based metric. PropaGator 2 produced a maximum of 56 𝑙𝑏𝑓 of thrust in a static thrust test where the ASV pulled on a scale that was tethered to the dock. The next strongest ASV was only able to produce 17 𝑙𝑏𝑓 of thrust.

7. CONCLUSION In this paper, the research and testing that went into the

mechanical design and implementation of PropaGator 2 has been described. A description of the design and shortcomings of PropaGator 1 lead to the motivation for developing an ASV capable of travelling 10 𝑘𝑡𝑠. The hydrofoil research and drag testing motivated the planing hull design used for PropaGator 2. The design of the various components of the steering and propulsion system was then detailed. The performance of the ASV was measured and was found to have exceeded all of the original design criteria.

Future development goals include redesigned motor pods with a motor built directly around the propeller inside of the Kort nozzle shroud. Some of the advantages of this new

Figure 8. Hubless propeller designed and manufactured for PropaGator 2

Table I. Propeller Parameters


thruster are that it will allow the water to act as a natural heat sink for the motors and it will eliminate the chance for a timing belt to loosen or break. The new thruster will also feature fewer and easier to access parts, which will require less maintenance than the current thrusters. Additionally, further empirical testing will be performed to improve the design of the propellers for the ASV. Finally, the ASV will be outfitted with radio frequency (RF) transmitters and receivers so that it can coordinate missions with other ASVs or an unmanned aerial vehicle (UAV).

8. ACKNOWLEDGEMENTS The authors of this paper would like to thank everyone

who has supported this project, including the University of Florida’s Electrical and Computer Engineering Department and Mechanical and Aerospace Engineering Department. The authors would like to extend an appreciative thank you to our adviser, Dr. Eric Schwartz (with whom this project was made possible), Dr. Antonio Arroyo, and the Machine Intelligence Laboratory at the University of Florida. Finally, the authors would like to thank Shannon Ridgeway and Jackson Graham who helped to make the production of a new hull a reality.

REFERENCES [1] Gray, A., Shahrestani, N., Frank, D., and Schwartz, E., 2013, “PropaGator 2013: UF Autonomous Surface Vehicle,” AUVSI Foundation’s 6th Annual RoboBoat Competition, Virginia Beach, VA. [2] Weaver, J., Frank, D., Schwartz, E., and Arroyo, A., 2013, “UAV Performing Autonomous Landing on USV Utilizing the Robot Operating System,” Proc. ASME District F - Early Career Technical Conference, Birmingham, AL, pp. 119–124. [3] Savitsky, D., 1985, “Planing Craft,” Naval Engineers

Journal, 97(2), pp. 113–141. [4] Eppler, R., 1990, Airfoil Design and Data. Springer-Verlag, Berlin, DE, Chap. 6. [5] Cengel, Y., and Cimbala, J., 2010, Fluid Mechanics : Fundamentals and Applications 2nd ed., McGraw-Hill Higher Education, New York, NY, Chap. 7. [6] Lamb, T., 2004, Ship Design and Construction Vol. II, Society of Naval Architects and Marine Engineers, Chap. 45. [7] Savitsky, D., 1964, “Hydrodynamic Design of Planing Hulls,” Marine Technology, 1(1), pp. 71–95. [8] Patel, D., Frank, D., and Crane, C., 2014 “Controlling an Overactuated Vehicle with Application to an Autonomous Surface Vehicle Utilizing Azimuth Thrusters,” Proc. 14th International Conference on Control, Automation, and Systems, Gyeonggi-do, KR, accepted. [9] Hsieh, M.-F., Chen, J.-H., Yeh, Y.-H., Lee, C.-L., Chen, P.-H., Hsu, Y.-C., and Chen, Y.-H., 2007, “Integrated Design and Realization of a Hubless Rim-Driven Thruster,” Proc. IECON 2007, 33rd Annual Conference of IEEE Industrial Electronics, Taipei, Taiwan, pp. 3033–3038. [10] Carlton, J., 2007, Marine Propellers and Propulsion 2nd ed., Butterworth-Heinemann, Oxford, UK, Chap. 2. [11] Yakovlev, A. Y., Sokolov, M. A., and Marinich, N. V., 2011, “Numerical Design and Experimental Verification of a Rim-Driven Thruster,” Proc. 2nd International Symposium on Marine Propulsors, Hamburg, DE, pp. 396–402. [12] Cao, Q.-M., Hong, F.-W., Tang, D.-H., Hu, F.-L., and Lu, L.-Z., 2012, “Prediction of Loading Distribution and Hydrodynamic Measurements for Propeller Blades in a Rim Driven Thruster,” Journal of Hydrodynamics, 24(1), pp. 50 – 57. [13] Gerr, D., 1989, Propeller Handbook, International Marine, Camden, ME, Chap. 5.



DESIGN RATIONALE FOR SUBZERO

Bennett R Stedwell, Ada A Odey, Kevin McFall PhD

Southern Polytechnic State University Marietta, GA, USA

ABSTRACT Southern Polytechnic State University’s SubZero is a

littoral-class autonomous underwater vehicle (AUV) built by undergraduate members of the SPSU AUV Team. The vehicle has been continuously modified and enhanced over the past several years, however the current configuration is a complete redesign from past vehicles, the product of a ten-month development cycle. The vehicle was designed almost completely using three-dimensional CAD and simulation in Dassault Systemes’ Solidworks design software. Among the new design’s features are redesigned camera housings, a new main housing, and a hexagonal, riveted 6061 aluminum exoskeleton to provide structural security while minimizing weight. SubZero is equipped with two cameras for challenge recognition and maneuvering computer vision tasks, a pressure sensor for active depth control, and an inertial measurement unit for orientation control.

Figure 1. SubZero

OPERATING SYSTEM AND LANGUAGES To avoid unnecessary complication, a software stack is

used to provide communications and interfacing with the sensors and cameras. This allowed for additional time to be spent addressing the challenges rather than perfecting the utilities. These functions were provided by Robot Operating System (ROS). As explained on ROS’s website, the Robot Operating System is a set of software libraries and tools that

assist in building robot applications. ROS possesses open source drivers, state-of-the-art algorithms, and other powerful developer tools. ROS has a six month release cycle; throughout the development cycle, new versions were released and thus had to be updated on SubZero, Figure 1. Fortunately, little code had to be rewritten between each update. Using ROS narrowed down the choices of operating system for the on-board embedded computer (ePC). ROS can be run on a variety of Linux distributions and Windows, however only one distribution is fully supported; Ubuntu. Code is not developed on the embedded PC (ePC); instead personal laptops or lab computers are used, which have ROS installed, to develop any software for SubZero. One reason for this decision was the concern that doing many compiles on the ePC could wear out the flash memory on the compact flash (CF) card, which is its only means of storage. It was decided to run Ubuntu on both the development and on-board computers, but with each installation customized for its purpose. The development installation has a desktop environment (Cinnamon), a browser, many editors, and other useful tools. The embedded PC installation boots directly into the terminal with the option to start a graphic user interface (GUI) for testing, and little extra software aside from ROS and the code, all of which must fit on an 8 gigabyte compact flash (CF) card. The ROS libraries in use are C++ and Python. Initially, it was decided to write in C++ for the speed of using native code, as the operating system itself is written in C++. After additional consideration, the code was switched to Python to maximize performance capabilities when vision processing tasks are running. When the code was ported to Python, there was no evidence of any performance lost on the other systems.

SOFTWARE ARCHITECHTURE A closed layer approach is used to design the ROS

package. This architecture was chosen for its ease of comprehension for those students new to computer programming. The four layers in the architecture are: Control, Decision, Calculating, and Device. Figure 2 shows the premise of the software stack. The nodes in each layer can only interact with nodes from the layers directly above or below, as well as nodes from its own layer.


CONTROL LAYER The control layer houses only one node: Intelligal

Artificialence (IA). This node’s purpose is to run the mission objectives that are defined in the decision layer.

DECISION LAYER The decision layer is where SubZero’s task decisions are

made. Each task has its own node in this layer; including the intermediate tasks such as re-covering from the previous activity. This approach simplifies updating code and streamlines the review and merging process in the content management system.

CALCULATING LAYER This layer handles all of the heavy number crunching

functions, such as locating the start gate in a frame. Located here are the visual recognition code, the nodes for collision detection, and the swim computer.

DEVICE LAYER This layer contains software from the ROS community that

pulls the data from the hardware and converts it into a usable

format. The device layer contains the rosserial nodes used to communicate with the two Arduino Unos, one to control the six thrusters and the other to read data from various sensors.

SWIM COMPUTER After consulting with colleagues on SPSU’s Aerial

Robotics Team (ART), they suggested the use of a “swim computer”. Unmanned Aerial Vehicles (UAVs) have an on board flight computer which is used to stabilize the aircraft. ART recommended that an actual flight computer board be placed in the vehicle, which could then be sent basic commands to handle stabilization and correcting for drift. However time constraints proved prohibitive to implementing this change, so “swim computer code” was instead built into the software.

CAMERA SENSOR SUITE The camera housings are constructed from 3” PVC piping,

two PVC slip fittings, two PVC slip to threaded fittings, two threaded PVC end caps, and finally two Lexan lenses. For a single housing, the pipe is cut to a length of 8.89 cm, a slip fitting and a slip to threaded fitting is glued in place using PVC glue. While the glue sets, the lens is cut to a diameter of 8.80 cm. Once the glue is fully cured, the lens is placed in the open end of the slip fitting and held in place with silicon.

The internal structure of SubZero's camera housings are constructed from 0.47 cm thick polyethylene, and two 90° steel elbow joints. Two pieces of polyethylene are cut using the same method as the Lexan lens; one is cut with an outside diameter of 8.80 cm and an inside diameter of 3.00 cm. The second piece is a 5.08 x 7.62 cm rectangle. The two pieces are connected using the elbow joints and the circular segment is epoxied inside the threaded end cap. The last three steps are: securing the USB camera to the rectangular section using double sided tape, inserting a Fischer connector into the back of the end cap, and finally attaching the USB wires to the connector; which runs between the ePC inside the main housing and the mounted camera housing. These steps are exactly repeated for the second camera housing.

The housings are attached to the exoskeleton so that one faces forwards and the other faces downwards. The downward camera is used to find the direction markers on the bottom of the pool which leads between the various competition tasks. The forward camera is used for computer vision targets such as the start gate.

For much of the vision code development cycle, FireWire cameras were used. These FireWire cameras have benefits and drawbacks. They do yield images at a high resolution. However, the cameras would randomly return a distorted image; the image would be divided in half and switched (right side on the left and vice versa), one half inverted and in gray scale, with the other half oriented correctly and in proper color. Since SubZero relies so heavily on a video feed from the cameras, it was decided to switch to a pair of Logitech USB web cams. This switch resolved all image distortion issues.

Figure 2. Software Architecture


Figure 3. Pressure Sensor Suite

Figure 4. Pressure Sensor Suite Pressure Simulation

DESIGNED MAIN HOUSING At the beginning of the year the mechanical lead

demonstrated a sketch which showed how a cylindrical housing would be able to hold all required electrical components in the current main housing. The cylindrical housing consists of clear PVC tube and two machined aluminum end caps. The end caps are designed in SolidWorks so that they can be made by a computer numerical control (CNC) machine at a very high level of precision. Aluminum was chosen for the end caps for its relatively low weight in comparison to other metals, as well as the high rate of heat transfer it would provide to the surrounding water. To take advantage of this the ePC is mounted directly to one of the end caps. The other end cap is designed to allow for multiple water-proof Fischer connectors to pass and connect to the electronics inside the housing. There are a total of 16 waterproof connectors intended to pass through the end cap into the housing. The two end caps utilize an O-ring channel so that the O-rings will be able to compress against the side wall of the tube, forming the watertight seal. The advantage of designing the parts in SolidWorks was that it became possible to utilize the simulation and analysis suites available to simulate the maximum pressure that the new main housing could sustain before being destroyed.

PRESSURE SENSOR SUITE In order for SubZero to perceive its own depth, it needs to

know the water pressure surrounding it. To do this, SubZero uses an Omega PX26 Series Pressure Transducer wired to the Adruino Uno. The pressure transducer is encased in its own housing to reduce the risk of a complete vehicle flooding. Since the pressure transducer has to be exposed to the surrounding water, there is a higher chance of the housing flooding. The housing, Figure 3, is milled from a 5.71 x 6.98 x 8.89 cm aluminum billet. The housing has two holes; one for the water side of the sensor, and the other for a four pin Fischer connector. This connector leads back to the main housing, and in turn the Arduino Uno; again passing through a four pin Fischer connector. While the housing is designed to handle depths of 18 m, Figure 4, it has only been tested to 3.66 m.

INERTIAL MEASUREMENT UNIT In addition to the pressure sensor and cameras, there is an

Inertial Measurement Unit (IMU). The IMU allows the vehicle to its current orientation, velocity, acceleration, and compass bearing. The orientation is based around rotation in the three primary axes and translation in the primary planes: Pitch, Roll, and Yaw, and dead reckoning in the planar motion. s of 18 m, Figure 4, it has only been tested to 3.66 m.

INERTIAL BASED NAVIGATION Inertial based navigation is a tool that appears to have seen

a recent increase in implementation due to recent advances in micro-electromechanical systems (MEMS) technology. The basic principles of its operation are that it uses a combination of three accelerometers and three rate gyroscopes oriented orthogonally to each other, to measure linear and angular accelerations in the three reference frame axes, say X,Y, Z although these letters are completely arbitrary. Given that these micro-sensors effectively measure the effects of forces, which by Newton’s second law is a mass times its acceleration, it follows that the accelerations may be integrated once to obtain velocity data, and twice to obtain position data. If a reference position and orientation is known, the IMU will output changes to this position(ΔV) and orientation(Δθ) at some fixed sampling rate in seconds, which may be then used by control software to ascertain how far the vehicle moved from the and how its orientation has changed from the reference orientation. The error in accelerometer and angular rate measurements being computed by the IMU manifests itself as a position drift proportional to the square of time (NEST, 2007). This error may be reduced by combining IMU data with other instruments, although SubZero’s missions are such that this is not necessary.

EXOSKELETON A significant factor in the creation of SubZero’s body

structure, Figure 5, is the fact that it is almost completely built from the ground up from scratch. The first step was to gather information and brainstorm ideas; research was conducted into the standard types of raw materials that are available for


immediate purchase on the market. A local home improvement and construction retailer became the source for many of these materials. This option proved expensive, so an alternate source was sought to provide bulk materials at a discount for a student competition team. A supplier of this kind was not found until later in the competition season. With a good idea of the types of raw materials available, each member of the team was tasked with submitting a rough exoskeleton structure design for the competition. Two of the best submitted designs were combined into a hybrid design combined the best features of both; a standing hexagonal prism design with mounting points for hardware components and housings. Once the structural design had been finalized, the next step was to order the tools and materials to construct a prototype. The prototype exoskeleton allowed the team to see flaws and improve the efficiency of the manufacturing process. For example, it was discovered that brake press bending machines are more accurate and better suited to preform bends, while a Chinese pipe bending press can perform larger-degree bends.

Figure 5. Exoskeleton

The team is permitted access by the school to a small milling machine, this allowed for the both time consuming and accurate placement of each mounting hole on the structure, followed by the four hole patterned joints. When this joint was tested it was found to be unnecessarily robust, and so a two- hole pattern was adopted. It was deemed necessary to build a prototype to validate the structural integrity of the design change, so a single ring was fabricated to perform deformation and impact testing which proved successful. With a solidified manufacturing plan, materials, and practice with the new equipment, the team proceeded to final exoskeleton build. Three types of raw materials are used to construct the final exoskeleton. 1” x 1/16” thick aluminum bar stock, selected for

use as light weight housing support structure, possesses the advantage of being inexpensive and easy to form. This is used to secure the camera housings and motors to the main frame. The three hexagonal structural rings are comprised of 1” x 1/8” bar stock. It was calculated that the increase in thickness of the metal provided an increase in stiffness, preventing excessive deflection while under the load of the components. The lower T-beam, of identical thickness to the hexagonal rings, formed the crucial lower structure for the frame and permitted the attachment of lighter supports directly to it, including metal “feet” to support the entire exoskeleton upright on a table. An advantage to the current exoskeleton frame is the versatility to add additional components with minimal changes in weight; it is simple to change component configurations, and the frame itself weighs seven pounds. The use of aluminum rivets was a topic of debate among the team, with good reasons for and against their use. They are lighter and tougher than a bolt of the same material, while being lower in cost. However, compared to bolts which need only washers, lock washers and bolts to be a fastener, rivets require specialized tools and training to properly install. It was decided that the structural benefits outweighed the costs of learning to use and properly install rivets, and so they were implemented as SubZero’s primary exoskeleton fasteners.

BALLAST TANKS The primary design criterion for the ballast tanks is to be

inexpensive and easy to build; Polyvinyl chloride (PVC) fits both of these requirements. To build the tanks, two 3” Polyvinyl chloride (PVC) pipes are cut to a length of 78.74 cm and a 3” PVC end cap are glued to the ends of each pipe. This completely seals the tanks against water leaking into the tanks. There is also a third tank made from a 2” PVC pipe and end caps, which has a length of 82.23 cm. The combined positive buoyancy from these three tanks is enough to counteract the weight of the entire vehicle and remain positively buoyant. All three tanks are attached to SubZero's frame using two standard 4” steel hose clamps per tank. As a safety feature, the vehicle is positively buoyant so that it floats to the surface in the event of a power failure. It was through trial and error that the correct pipe lengths were found.

While the current ballast tanks work for the current vehicle, the next version will not use PVC piping, but instead use two thin walled aluminum pipes. The next iteration will be lighter weight, function at a greater depth, and have smaller over all dimensions while still supporting the same amount of weight.

ELECTRICAL SAFETY The Kill Switch is a magnetic reed switch similar to a door sensor for home security systems. When a magnet is brought close enough, the switch will change states. The switch is hooked up to the second Arduino Uno and a 30 A relay. When the switch changes state, the Arduino sends a signal to the ePC informing the vehicle of which state the motors are now in. The


relay is connected between the thruster batteries and the motor drivers, thus when the switch is in the off position, no current can pass through the relay.

A magnetic switch was chosen in order to reduce the number of possible places the vehicle could leak. The magnet is attached to a blaze orange float for the safety diver to easily grab in the event of an emergency. Pulling the float will remove the magnet, which cuts the current to the motor drivers and putting the vehicle in a safe state for the diver to handle.

ELECTRICAL POWER DISTRIBUTION Initially the power system was housed separately from

all the electronics; however this was found to significantly increase the weight of SubZero but had little additional benefit. The current fully integrated setup is simple and organized, with the focus of keeping power system maintenance as easy as possible. Motor power is provided by two 14.8 V, 5 AH lithium-polymer batteries connected in parallel. The maximum voltage rating on the motors is 19 V, thus a parallel setup is used to stay under this limit while benefiting from an increased mission time. The power from the batteries moves through the kill switch, and then gets passed on to three motor drivers to individually power the motors. The main power for the computers is a 12 V, 7 AH lithium-iron-phosphate motorcycle battery. Originally all power was to be stepped down from the motor battery setup; however this would have required additional circuitry and thus another point of possible failure. The lithium-iron-phosphate battery powers the ePC, which then powers the rest of the electronics: the Arduino's, IMU, Camera, etc.

PROPULSION The submarine utilizes six SeaBotix thrusters for

maneuverability. They are brushed DC motors encased in a waterproof housing. They are able to produce a peak thrust of 2.9 kg force and are controlled by three Sabertooth motor drivers that communicate by simple serial. These drivers give users the ability to control the rotational speed and direction of the thrusters.

SUBZERO’S PERFORMANCE The maximum depth achieved by SubZero is 3.66 m,

which is the deepest part of the test pool. The maximum depth of the camera housings, which are the weakest components, is unknown because the housing is a replication of a field proven idea.

SubZero’s vision processing is capable of finding two parallel vertical orange bars floating in the water and then guiding the vehicle towards the center point between the bars.

PLANS FOR NEXT YEAR Currently each task is split in multiple files; the main logic

has its own node, but any supporting logic, such as vision, is in another node in the calculating layer. In order to streamline the code, everything is placed in one file but have the supporting nodes import any necessary functions from the main node. Another plan is to change the ePC to a Hardkernal ODroid U3, which is the same computer ART uses. The U3 has comparable performance specifications as the current ePC, but with a significantly smaller footprint and electrical power draw. A major drawback to the U3 is it only has 2GB of RAM instead of the 4 GB found on the current ePC. Finally, an off the shelf RC flight computer will be placed in SubZero to handle vehicle drift and orientation.

Another step forward will be to finish the designed housing. While the current one works, the designed housing is significantly lighter and more compact which will allow for a short vehicle.

Finally, three things need to be done with the camera housings: CAD modeling, computer finite element analysis (stress testing) and finally field testing to verify the computer simulations.

ACKNOWLEDGMENTS We would like to thank our former academic adviser,

Professor Scott Tippens, for putting up with us throughout the year. It has been a rather chaotic year for the team, but he has helped us make as far as we have. We would also like to acknowledge the rest of our team: Douglas Allyson, Albert Cheng, Alex Evans, Andrew Geiman, Lorenza Hill, Taylor Martin, and Nikhil Ollukaren. Without them, we would not have been able to finish our vehicle. Additionally, we would like to acknowledge the SPSU Student Government Association and Alumni Foundation for giving us our budget for the year. Finally, we would like to thank Marietta Diver Supply Co., who allowed us to use their pool facilities and SCUBA equipment for the vehicle testing.

REFERENCES [1] Naval Engineering Support Team (NEST) (2007). AUVSI/ONR Engineering Primer Document for the Autonomous Underwater Vehicle (AUV) Team Competition. AUVSI Foundation. Retrieved October 2, 2014, from http://higherlogicdownload.s3.amazonaws.com/AUVSI/fb9a8da0-2ac8-42d1-a11e-d58c1e158347/UploadedImages/Support_Primer_r1.pdf



THERMAL ANALYSIS, MICROSTRUCTURAL CHARACTERIZATION AND NANOINDENTATION FOR ELECTRON BEAM ADDITIVE MANUFACTURING

Bo Cheng, Xibing Gong, Xiaoqing Wang and Kevin Chou Mechanical Engineering Department

The University of Alabama Tuscaloosa, Alabama, USA

ABSTRACT In a current research project focused on electron beam

additive manufacturing (EBAM) using Ti-6Al-4V material, several critical issues related to the process temperatures and part properties have been studied including thermal responses from finite element model, build part microstructures and mechanical properties analysis. Finite element numerical analysis was applied for EBAM process temperature simulations. Build part microstructures were studied by optical and scanning electronmicroscopy. Moreover, mechanical properties of the build parts have been investigated by nanoindentation technology.

Numerical simulation is capable of predicting temperature distributions during the electron beam melting process. X-plane (side surface) specimens had columnar prior β grains with martensitic structures, while the Z-plane (scanning surface) specimens showed equiaxed grains. The obtained elastic modulus values of X-plane and Z-plane are 112 GPa and 116 GPa while hardness values are 6.0 GPa and 5.87 GPa, respectively.

INTRODUCTION The electron beam additive manufacturing (EBAM)

technology utilizes high-energy sources to fabricate metallic parts, in a layer by layer fashion, by sintering and/or melting metal powders. EBAM offers a unique combination of process flexibility, time savings and reduced cost in producing complicated components. There are three major steps which include powder spreading, pre-heating, and contour/hatch melting during the layer building process. In the powder spreading step, a metal rake is utilized to distribute one layer of powder with a given layer thickness on the previously deposited material. In the preheating step, a high-energy electron beam is used to do multi-pass scans at a high speed and beam current (e.g., ~15 m/sec and ~30 mA), to slightly sinter the top powder layer. This process is followed by contour-melting and hatch-melting in a selective region defined by 3D model, CAD data. Any desired part can be built with such repeated cycles. The EBAM process has the capability of

making fully-dense, functional metallic parts with product material properties comparable or even better than parts made by conventional means. The build parts are also free from serious distortions since the process is under a vacuum at elevated temperatures [1-3].

There are still some challenges and difficulties for industrial widespread applications of EBAM, despite the listed advantages and potential benefits. In this EBAM study, three technical aspects have been investigated with methodology and results discussed. These are process thermal responses from finite element model, build part microstructure and mechanical properties analysis. The final goal is to develop correlations between thermal process modeling, microstructures and mechanical properties of EBAM parts for quality control.

Numerical simulation of EBAM process is a challenging task due to the complex physical processes introduced and the interactions among thermal, mechanical, and metallurgical phenomena. Mahale [4] applied finite element methods and finite difference methods to develop a 3D model by defining a scan path into COMSOL Multiphysics using Matlab for simulating of the EBM process for Al 7075. Planar heat source as well as temperature dependent material properties were employed with any heat loss due to radiation being ignored in the model. A variable meshing method was also used to achieve higher resolution around the beam scanning region.

Zäh and Lutzmann [5] conducted a thermal analysis of EBAM with FEA, taking into account the metallic powder in the powder layer itself. The effects of electron beam power and scanning speed on the molten pool geometry were also studied. Jamshidinia et al [6] developed a thermal-fluid flow model of the electron beam melting process incorporating powder thermal properties to study the influence of process parameters such as electron beam scanning speed, electron beam current, and the powder bed density. The effects of flow convection in temperature distribution and molten pool geometry were also investigated.

Microstructures of Ti-6Al-4V samples from EBAM contain a mixture of phases such as α, β and α´ martensite. The columnar prior β structure formed during initial solidification has been observed, which is a result of very high temperature


gradients, along the build direction [7, 8]. Antonysamy et al. [9] studied the prior β grain texture of EBAM components and reported that the columnar structure shows strong fiber texture of <001> β, which is normal to the deposited powder layers. Safdar et al. [8] reported typical Widmanstätten (α + β) inside the prior β grains. Facchini et al. [10] investigated microstructures and showed that the main constituent is α with only a small fraction of β. Christensen et al. [11] compared the EBAM microstructures to counterparts from cast specimens. For the cast specimen, a coarse acicular (α + β) and thick prior β grain boundaries can be observed, while the microstructure of the EBAM specimen consisted of fine acicular (α + β) and thin prior β grain boundaries.

The thickness of the α-lath is around 1.4-2.1 µm for different EBAM samples [12]. Due to high cooling rates during the solidification, α´-martensitic platelets exist in EBAM parts, which may contribute to increased strength and hardness but lower ductility [3, 13]. In summary, Ti-6Al-4V samples from EBAM show a fine Widmanstätten (α + β) microstructure combined with α´, which can be expected by the thermal characteristics of the EBAM processhaving a small melt pool and rapid cooling. However, the microstructures of EBAM Ti-6Al-4V can vary with changes to processing conditions.

The mechanical properties of Ti-6Al-4V have been widely reported in literatures. Facchini et al. [10] has obtained an elastic modulus for Ti-6Al-4V of 118 GPa with a standard deviation of 5 GPa by tensile test. On the other hand, results from Baufeld and Van der Biest [14], have the elastic modulus of Ti-6Al-4V at 117 GPa by tensile test with impulse excitation being independent of orientation and location. The elastic modulus for the wrought or cast Ti-6Al-4V was around 120~125 GPa with hardness around 4~4.2 GPa [15, 16]. Murr et al. [17] measured micro-indentation hardness of a specimen and found it to be about 3.8~4.1 GPa, which was very close to wrought alloy hardness.

Nanoindentation is a non-destructive technique used for evaluating mechanical properties from a very small volume of material by deforming it with an indenter. This technique has been widely used to study mechanical behavior of Ti-6Al-4V alloys [15, 16, 18]. Unfortunately, there have been very few studies on the EBAM-processed Ti-6Al-4V alloy.

In this research, a finite element model, incorporating a high energy moving heat source, temperature dependent material properties, latent heat of fusion, etc., was used to evaluate the thermal response of a EBAM process. Optical microscope and scanning electron microscope were utilized to analyze build part microstructural characteristics. In addition, nanoindentation tests have been conducted to acquire part mechanical properties such as elastic modulus and hardness.

EXPERIMENTS An Arcam S12 EBAM machine, at NASA’s Marshall

Space Flight Center (Huntsville, AL), was used to fabricate samples. Samples were simple blocks (60 mm long, 5.5 mm

wide and 25 mm high) modeled from CAD software. Fine pre-alloyed Ti-6Al-4V powder, with a diameter between 45 and 100 µm, was used as feedstock material. The machine default settings typically employed for Ti-6Al-4V powders were used during the fabrication process. During the part fabrication process, a set of parameters programmed into the machine operating system, known as speed functions, were used to automatically control the electron beam scanning speed and current. In addition, the electron beam diameter was controlled by beam focus offset [19]. The layer thickness for part building was set at 0.07 mm. Figure 1 shows the Arcam S12 EBAM machine used in the experiment.

Figure 1. Arcam EBAM Machine: (a) Overall Appearance, (b) Inside of Build Chamber

TECHNICAL STUDIES Finite Element Thermal Simulation

A 3D FE thermal simulation model was developed using commercial software (ABAQUS), to study the complex thermal phenomenon in EBAM. A deposited top layer is considered as the newly added powder material while previously deposited layers are modeled as solid material. The electron beam heating occurs at the top surface of the powder layer and travels along the x-axis with a constant speed. To reduce computational time, a fine mesh is applied along the scanning path of the electron beam which directly incorporating the incident Gaussian heat-flux region, while the regions away from main heat affected


zone used coarse mesh. In addition, since a single path scan was simulated in this study, only half of the domain has been modeled. To simplify the physical process of electron beam travel, the elements/nodes remained inside the mesh even when temperature exceeds the evaporating point of the material. Since the EBAM process is conducted in a vacuum, any convection between the model and environment is not considered, only the thermal radiation is considered between the model and surroundings. A uniform temperature, Tpreheat, is assigned to the whole model to simulate the initial preheating effect. During the electron beam scanning process, the bottom surface of the model has been confined to a fixed temperature, which is also Tpreheat, to simulate the thermal boundary condition. Figure 2 shows the model configuration and boundary conditions.

Figure 2. Model Configuration and Boundary Conditions

To incorporate a moving heat source, a user defined

subroutine, DFLUX, was developed in FORTRON. It reads simulation time step and model coordinates to determine the domain of the volumetric heat flux with Gaussian distribution. Another user subroutine, UMATHT, was also developed to assign material properties according to material state changes, e.g., powder to liquid, and then to solid. An index is defined to describe material state, 0 for powder, 1 for solid. The change criterion is molten material under cooling (i.e. T > Tm & dT/dt < 0).

Widely used Ti-6Al-4V powder was used in this study. Temperature-dependent material properties were incorporated into the developed model. Temperature dependent properties of solid Ti-6Al-4V alloy have been reported such as density [20, 21], and specific heat [22]. In particular, the thermal conductivity of Ti-6Al-4V solid and powder were experimentally measured from build samples [23]. The linear interpolation method was used to obtain conductivity values at

different temperatures for both solid and powder. The increase of thermal conductivity after melting point was used to account for molten pool convection phenomenon according to Liu et al. [24]. Other process parameters, including powder layer thickness, electron beam acceleration voltage, and preheat temperature, etc., are all listed in Table 1.

Table 1. Process Parameters Used in Simulation

Parameters Value Solidus temperature, TS (°C) 1605 [25] Liquidus temperature, TL (°C) 1655 [25] Latent heat of fusion, Lf (kJ/Kg) 440 [25] Electron beam diameter, Φ (mm) Experiment Absorption efficiency, η 0.9 [26] Scan speed, v (mm/sec) Experiment Acceleration voltage, U (kV) 60 [27] Beam current, Ib (mA) Experiment Powder layer thickness, tlayer (mm) 0.07 Porosity, φ 0.5 Beam penetration depth, dP (mm) 0.062 [5] Preheat temperature, Tpreheat (°C) 730

One example of actual process parameters in the

experiments, at a build height of around 6.37 mm, has been incorporated into the FE model (991 mm/s scanning speed, 9.2 mA beam current and approximately 0.55 mm beam diameter). A single straight scan simulation was conducted and the temperature profile along the scanning path was extracted for further analysis. Figure 3 shows thermal responses in the EBAM process. Figure 3 (a) shows the temperature contours when the melt pool reached steady state. The melt pool shape is shown by turning regions where temperature is under the liquidus point to white color. The melt pool shows a typical moving style with a long tail in the opposite direction of electron beam scanning. In addition, the melt pool is around 3.45 mm in length, 1.09 mm in width and 0.104 mm in depth. Figure 3 (b) illustrates the temperature profile extracted from the center line of the scan path with the beam center at 0 on the horizontal axis. Rapid temperature rise and fall near the beam center are noted, with the peak temperature reaching 2937 °C. The plateau area is noted since the latent heat of fusion for the phase change is considered in the model.

Grain growths, crack nucleation and residual stresses may be affected by cooling rate in melt pool. Therefore, heating/cooling rates are also investigated in this study. Figure 3 (c) shows the temperature heating and cooling history at one


location in the scan path close to the starting point. The plateau is noted again at around 2 µs, the cooling rate drops to a very low value during this period. The phase change event can be noted by the abrupt drop in heating period. In addition, the change from the latent heat effect to solid-state cooling can be seen by another sharp transition that occurred at around 6.8 ms.

Figure 3. Temperature Responses of Simulated Case

Microstructure Study The fabricated Ti-6AL-4V samples were prepared for

microstructural observations with standard metallographic procedures including sectioning, mounting, grinding with SiC papers up to a grit size of 1000, and then polished using diamond suspension down to 0.5 µm. Specimens close to the

top surface of the build parts were used for analysis. Specimens of different cross-sections (scanning surface: Z-plane, build side surface: X-plane) were prepared to examine the anisotropic conditions in microstructures. To reveal the microstructures, polished specimens were then etched with a hydrofluoric acid-based solution (1 mL hydrofluoric acid (50 wt. %) and 3 ml (60 wt. %) nitric acid in 7 ml distilled water). The etched metallographic samples were examined using a Leitz optical microscope (OM) and a Philips XL-30 scanning electron microscope (SEM). In order to quantify the size of columnar β, equiaxed β and α-lath, a measurement method, from Wang et al. [28], was applied.

Figure 4 shows typical microstructures from the X-plane (side surface) of the EBAM sample. It is obvious that the prior β grains grew along the build direction and across multiple layers in the sample. The solidification of Ti-6Al-4V alloy involves a transformation of liquid to a primary solid phase of β and, then, a solid phase transformation (β to α or α´) that is dependent upon the cooling rate. The nucleation and growth of prior β columnar grains takes place during initial rapid solidification when the temperature is above the β-transus temperature (about 980 °C). The prior β columnar grains are typical in high-energy material processing, and upon rapid cooling from the melt, these growing grains align themselves with the steepest temperature gradients [29]. This results in a columnar shaped morphology [30]. According to Antonysamy et al. [9], nucleation of β grains occurred heterogeneously from boundary layers at the build plate or part surfaces. The authors also studied the prior β-grain texture of EBAM components and the columnar structure shows strong fiber texture of <001> β along the build direction, which could be attributed to the elongated shape of the moving melt pool.

Another feature seen in the EBAM Ti-6Al-4V microstructures is the martensitic phase, α´, which appears as plates, in Figure 4. The α´ transformed from the β phase due to a very high cooling rate. According to Ahmed and Rack [31], for Ti-6Al-4V alloys, a cooling rate greater than 410 °C/s, from the single β to the α + β region, will induce α´ formation. In addition, to trigger the formation of martensite, the temperature must be lower than the martensite starting temperature (MS) of 575 °C [29]. Other researchers, however, reported the MS as 650 °C and the martensite finishing temperature as 400 °C [32].

The width of the prior β columnar grains was measured in this study. Figure 4 (a) illustrates an example of the width measurements. The image is overlaid with straight lines which are normal to the boundaries of the columnar structures and the intersections of the lines with grain boundaries were examined and used for width estimates. Different images from different specimen areas were measured to obtain statistical data. The average width of the columnar structure is about 41.6 µm for this sample. This measurement is smaller than the columnar width reported by Al-Bermani et al. [7], 75 to 150 µm.


(b)

α′

Figure 4. Microstructure of X-Plane (Side Surface) Specimen from EBAM Sample: (a) Low Magnification

Showing Columnar Width Analysis, and (b) High Magnification Image

Figure 5 and Figure 6 show typical microstructures

consisting of equiaxed grains in the Z-plane (scanning surface) specimen. This is different from the microstructure in the X-plane (see Figure 4). It can be concluded that the prior β grains are of a rod shape. Similar equiaxed grains at the scanning surface has also been reported by other researchers for Ti-6Al-4V alloy [7] and other similar alloys [33, 34]. Figure 5 (b) and Figure 6 show higher magnification images of the Z-plane specimen where both α and β phases can be identified. Upon cooling from the β-transus temperature, the initial α that nucleates is “grain boundary” α (αGB) with a location on a β boundary. Eventually, the β boundaries will be replaced by αGB in a continuous fashion [29], as can be noted from Figure 6 (a). In addition, fine Widmanstätten (α + β) structures are shown inside of equiaxed grains, indicating a rapid cooling rate in the EBAM process. Widmanstätten (α + β) is a typical microstructure of Ti-6Al-4V alloys produced by EBAM, as can

be clearly identified from high magnification images shown in Figure 5 (b) and Figure 6 (b) respectively. In solid phase transformations, the prior β columnar grains are transformed into fine α laths. The (α + β) structure is formed by diffusion controlled solid phase transformation, in which V diffuses into β while Al diffuses into α [8]. Furthermore, the classical α-lath structure is surrounded by a very small amount of β in the α boundaries. The result is similar to studies from Safdar et al. [8] and Murr et al [13]. Compared with the wrought or cast Ti-6Al-4V alloys, which shows coarse α laths or equiaxed α/β [12], EBAM processed Ti-6Al-4V parts show a finer α phase. Al-Bermani et al. [7] reported that fine α laths have no preferred texture which is different from strong texturing of prior β grains.

(a)

(b)

Figure 5. Microstructure of Z-Plane (Scanning Surface) Specimen from EBAM Sample: (a) Low Magnification

Showing Grain Size Analysis, and (b) High Magnification Image

Similar to the measurement method used for the columnar width, intersections of the lines with prior β grain boundaries


were marked and used for grain size estimates. The microstructural image was overlaid with several random lines, as shown in Figure 5 (a). From the measurements, an estimated size of the equiaxed grains is 50.1 µm. The width of α laths was also quantitatively analyzed using SEM images as shown in Figure 6 (b). The average thickness of α laths is 1.1 µm for this particular sample.

(a)

αGB

(b)

α′

β α

Figure 6. Scanning Electron Micrographs from Z-Plane of EBAM Sample: (a) 1000 × Image, and (b) 5000 × Image

Showing α-Lath Analysis

Mechanical Properties Study

The nanoindentation tests were conducted using Triboindenter with the Berkovich tip with a radius of 100 nm and an included angle of 142.3°. The maximum load is 10.0 mN with resolution of 1 nN, and the maximum depth is 20 μm with displacement resolution of 0.04 nm. An open-loop trapezoidal shape loading and unloading method with the maximum load of 5000 μN was applied to the sample during

nanoindentation testing. In trapezoidal shape method, the loading force on the indenter increases at a constant load rate until reaching a maximum value, then followed by a constant unload rate until reaching zero. To obtain the precise results, several nano-indenatation tests were conducted on different areas of the specimen. The indent pattern used is a matrix of 5 × 5 with spacing of 5 μm between any two horizontal or vertical adjacent points.

Nanoindentation tests were performed on the specimens by a 5 × 5 matrix with the same maximum load of 5000 µN. The depth vs. displacement values on 25 points has been obtained. With the plots of load vs. depth obtained from the tests, the elastic modulus of the sample can be calculated by:

1𝐸𝑟

=1− 𝜈𝑖2

𝐸𝑖+

1 − 𝜈𝑠2

𝐸𝑠

where νi=0.07 (Poisson's ratio of indenter), Ei=1140 GPa (elastic modulus of indenter), νs=0.342 (Poisson's ratio of Ti-6Al-4V). The hardness has the normal definition:

𝐻 =𝑃𝑚𝑎𝑥

𝐴 where Pmax is the maximum indentation force and A is the resultant projected contact area at that load evaluated from the shape function of the indenter and the maximum indent displacement.

The elastic modulus of the Ti-6Al-4V parts built by EBAM is 116.26 GPa on Z-plane (scanning surface) and 111.67 GPa on X-plane (side surface), and the hardness values on the Z-plane and X-plane are 5.87 GPa and 6.0 GPa, respectively. A comparison of the test results with literature data are shown in Figure 7 and Figure 8 respectively.

Figure 7. Elastic Properties of Ti-6Al-4V sample

0.00

20.00

40.00

60.00

80.00

100.00

120.00

Z-plane X-plane Facchini Baufeld

E (G

Pa)

[10] [14]


Figure 8. Hardness of Ti-6Al-4V sample

It can be observed that the elastic modulus has almost the same value on the Z-plane (scanning surface) and X-plane (side surface). Comparing these values to those obtained by other researchers, the elastic modulus of the sample built by EBAM is in fact comparable.

On the other hand, the EBAM Ti-6Al-4V sample has a slight higher hardness than that stated in the literature. This difference may be the result of a difference in the measuring method. According to the results of Qian et at.[35], the nanoindentation hardness is 10%~30% higher than the micro-hardness. Thus, the experimental values obtained by nanoindentation tests would in fact agree well with the literature data.

CONCLUSION In this study, Ti-6Al-4V samples built in EBAM have been

studied using numerical simulation, microstructural evaluations and nanoindentation experiments. A developed 3D thermal model was applied to investigate typical process parameters effect on part temperature distributions and melt pool sizes. Build sample microstructural characteristics and mechanical properties have been investigated through experiments. The major findings can be summarized as follows:

(1) The peak process temperatures in EBAM using Ti-6Al-4V powder are on the order of 2937 °C, with the melt pools having dimensions of about 3.45 mmx 1.09 mm x 0.104 mm (length x width x depth) , at a beam speed of 991 mm/s, a beam current of 9.2 mA, and a diameter of and 0.55 mm.

(2) The X-plane (side surface) specimens show prior β columnar grains with martensitic structures. The measured width of columnar grains is in the range of 41.6 µm. The Z-plane (scanning surface) specimens show equiaxed grains. The microstructure inside of the equiaxed grains is Widmanstätten (α + β).

(3) The elastic modulus of the EBAM Ti-6Al-4V parts is 116.26 GPa on Z-plane (scanning surface) and 111.67 GPa on X-plane (side surface). The hardness values on the Z-plane and

X-plane are 5.87 GPa and 6.0 GPa, respectively, which are in good agreement with literature results.

ACKNOWLEDGEMENTS The materials presented in this paper are supported by

NASA, under award No. NNX11AM11A. The research is in collaboration with the Marshall Space Flight Center, Advanced Manufacturing Team.

REFERENCES [1] Available from: http://www.arcam.com/. Accessed June 21, 2013. [2] Biamino, S., Penna, A., Ackelid, U., Sabbadini, S., Tassa, O., Fino, P., Pavese, M., Gennaro, P., and Badini, C., 2011, "Electron beam melting of Ti–48Al–2Cr–2Nb alloy: Microstructure and mechanical properties investigation," Intermetallics, 19(6), pp. 776-781. [3] Gong, X., Anderson, T., and Chou, K., 2014, "Review on powder-based electron beam additive manufacturing technology," Manufacturing Review, 1, pp. 1-12. [4] Mahale, T. R., 2009, "Electron beam melting of advanced materials and structures," Ph.D. thesis, North Carolina State University, Raleigh, NC. [5] Zäh, M. F., and Lutzmann, S., 2010, "Modelling and simulation of electron beam melting," Production Engeering. Research and Development, 4, pp. 15-23. [6] Jamshidinia, M., Kong, F., and Kovacevic, R., 2013, "Numerical Modeling of Heat Distribution in the Electron Beam Melting® of Ti-6Al-4V," Journal of Manufacturing Science and Engineering, 135(6), p. 061010. [7] Al-Bermani, S., Blackmore, M., Zhang, W., and Todd, I., 2010, "The origin of microstructural diversity, texture, and mechanical properties in electron beam melted Ti-6Al-4V," Metallurgical and Materials Transactions A, 41(13), pp. 3422-3434. [8] Safdar, A., Wei, L.-Y., Snis, A., and Lai, Z., 2012, "Evaluation of microstructural development in electron beam melted Ti-6Al-4V," Materials Characterization, 65, pp. 8-15. [9] Antonysamy, A., Meyer, J., and Prangnell, P., 2013, "Effect of build geometry on the β-grain structure and texture in additive manufacture of Ti6Al4V by selective electron beam melting," Materials Characterization, 84, pp. 153-168. [10] Facchini, L., Magalini, E., Robotti, P., and Molinari, A., 2009, "Microstructure and mechanical properties of Ti-6Al-4V produced by electron beam melting of pre-alloyed powders," Rapid Prototyping Journal, 15(3), pp. 171-178. [11] Christensen, A., Kircher, R., and Lippincott, A., "Qualification of electron beam melted (EBM) Ti6AI4V-ELI for orthopaedic applications," Proc. Medical Device Materials IV: Proceedings of the Materials and Processes for Medical Devices Conference, pp. 48-53. [12] Murr, L., Esquivel, E., Quinones, S., Gaytan, S., Lopez, M., Martinez, E., Medina, F., Hernandez, D., Martinez, E., and Martinez, J., 2009, "Microstructures and mechanical properties

0.00

1.00

2.00

3.00

4.00

5.00

6.00

Z-plane X-plane Murr

H (G

Pa)

[17]


http://www.arcam.com/�

of electron beam-rapid manufactured Ti–6Al–4V biomedical prototypes compared to wrought Ti–6Al–4V," Materials Characterization, 60(2), pp. 96-105. [13] Murr, L., Gaytan, S., Medina, F., Martinez, E., Martinez, J., Hernandez, D., Machado, B., Ramirez, D., and Wicker, R., 2010, "Characterization of Ti–6Al–4V open cellular foams fabricated by additive manufacturing using electron beam melting," Materials Science and Engineering: A, 527(7), pp. 1861-1868. [14] Baufeld, B., and Van der Biest, O., 2009, "Mechanical properties of Ti-6Al-4V specimens produced by shaped metal deposition," Science and technology of advanced materials, 10(1), p. 015008. [15] Xue, W., Wang, C., Chen, R., and Deng, Z., 2002, "Structure and properties characterization of ceramic coatings produced on Ti–6Al–4V alloy by microarc oxidation in aluminate solution," Materials Letters, 52(6), pp. 435-441. [16] Barbieri, F., Otani, C., Lepienski, C., Urruchi, W., Maciel, H., and Petraconi, G., 2002, "Nanoindentation study of Ti6Al4V alloy nitrided by low intensity plasma jet process," Vacuum, 67(3), pp. 457-461. [17] Murr, L. E., Quinones, S. A., Gaytan, S. M., Lopez, M. I., Rodela, A., Martinez, E. Y., Hernandez, D. H., Martinez, E., Medina, F., and Wicker, R. B., 2009, "Microstructure and mechanical behavior of Ti–6Al–4V produced by rapid-layer manufacturing, for biomedical applications," Journal of the Mechanical Behavior of Biomedical Materials, 2(1), pp. 20-32. [18] Karlsson, J., Snis, A., Engqvist, H., and Lausmaa, J., 2013, "Characterization and comparison of materials produced by Electron Beam Melting (EBM) of two different Ti–6Al–4V powder fractions," Journal of Materials Processing Technology, 213(12), pp. 2109-2118. [19] Gong, H., Rafi, K., Starr, T., and Stucker, B., 2013, "The Effects of Processing Parameters on Defect Regularity in Ti-6Al-4V Parts Fabricated By Selective Laser Melting and Electron Beam Melting," 24rd Annual International Solid Freeform Fabrication Symposium - An Additive Manufacturing Conference, Austin, TX, USA, August 12-14, 2013. [20] Yang, J., Sun, S., Brandt, M., and Yan, W., 2010, "Experimental investigation and 3D finite element prediction of the heat affected zone during laser assisted machining of Ti6Al4V alloy," Journal of Materials Processing Technology, 210(15), pp. 2215-2222. [21] Rai, R., 2008, "Modeling of heat transfer and fluid flow in keyhole mode welding," Ph.D. thesis, The Pennsylvania State University, State College, PA. [22] Mills, K. C., 2002, Recommended values of thermophysical properties for selected commercial alloys, Woodhead Publishing. [23] Gong, X., Cheng, B., Price, S., and Chou, K., 2013, "Powder-bed Electron-Beam-Melting Additive Manufacturing: Powder Characterization, Process Simulation and Metrology," ASME District F Early Career Technical Conference, Birmingham, AL, November 2-3, 2013.

[24] Liu, C., Wu, B., and Zhang, J., 2010, "Numerical Investigation of Residual Stress in Thick Titanium Alloy Plate Joined with Electron Beam Welding," Metallurgical and Materials Transactions B, 41(5), pp. 1129-1138. [25] Boyer, R., Welsch, G., and Collings, E. W., 1998, "Materials Properties Handbook: Titanium Alloys," ASM InternationalMaterials Park, OH, USA, pp. 483-636. [26] Rouquette, S., Guo, J., and Le Masson, P., 2007, "Estimation of the parameters of a Gaussian heat source by the Levenberg-Marquardt method: Application to the electron beam welding," International Journal of Thermal Sciences, 46(2), pp. 128-138. [27] Gaytan, S. M., Murr, L. E., Medina, F., Martinez, E., Lopez, M. I., and Wicker, R. B., 2009, "Advanced metal powder based manufacturing of complex components by electron beam melting," Materials Technology, 24, pp. 180-190. [28] Wang, K., Zeng, W., Shao, Y., Zhao, Y., and Zhou, Y., 2009, "Quantification of microstructural features in titanium alloys based on stereology," Rare Metal Materials and Engineering, 38(3), pp. 398-403. [29] Kelly, S. M., 2004, "Thermal and microstructure modeling of metal deposition processes with application to Ti-6Al-4V," Ph.D. Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, VA. [30] Wu, X., Liang, J., Mei, J., Mitchell, C., Goodwin, P., and Voice, W., 2004, "Microstructures of laser-deposited Ti–6Al–4V," Materials & design, 25(2), pp. 137-144. [31] Ahmed, T., and Rack, H., 1998, "Phase transformations during cooling in α+β titanium alloys," Materials Science and Engineering: A, 243(1), pp. 206-211. [32] Elmer, J., Palmer, T., Babu, S., Zhang, W., and DebRoy, T., 2004, "Phase transformation dynamics during welding of Ti–6Al–4V," Journal of applied physics, 95(12), pp. 8327-8339. [33] Amato, K., Hernandez, J., Murr, L., Martinez, E., Gaytan, S., Shindo, P., and Collins, S., 2012, "Comparison of Microstructures and Properties for a Ni-Base Superalloy (Alloy 625) Fabricated by Electron Beam Melting," Journal of Materials Science Research, 1(2), pp. 3-41. [34] Murr, L. E., Gaytan, S. M., Ceylan, A., Martinez, E., Martinez, J. L., Hernandez, D. H., Machado, B. I., Ramirez, D. A., Medina, F., Collins, S., and Wicker, R. B., 2010, "Characterization of titanium aluminide alloy components fabricated by additive manufacturing using electron beam melting," Acta Materialia, 58(5), pp. 1887-1894. [35] Qian, L., Li, M., Zhou, Z., Yang, H., and Shi, X., 2005, "Comparison of nano-indentation hardness to microhardness," Surface and Coatings Technology, 195(2–3), pp. 264-271.



Documents

UAB - ECTC 2014 PROCEEDINGS - Section 7 Page