DUKE ECO-MARATHON TEAM
The Duke Eco-marathon Team (DEMT) spends each academic year designing and manufacturing an ultra-efficient electric vehicle from the ground up. The group competes in the annual Shell Eco-marathon competition, in which teams are challenged to build the most energy efficient vehicle possible. During the competition, teams perform trial runs with their vehicle on a pre-determined test track and are given efficiency scores based on the energy consumed during each run. Team's are required to complete the course while maintaining an average speed of at least 15 mph.
In 2013, as a senior at Duke, I took over as President of the DEMT. The team was started in 2011, and had earned 10th and 3rd place at the 2012 and 2013 Shell Eco-marathon
competitions respectively. As President I took on the challenge of building upon our team's success while working to cultivate a learning environment in which young engineers could build valuable practical engineering skills. In doing so, I led our 15-member team through the design engineering journey outlined below.
Identify Key Parameters
We started our design process with a series of reflection sessions. Our sessions were focused around identifying the design weaknesses of our 2013 vehicle (pictured on the right), and led to the identification of three specific parameters that would become the main focus of our design process:
Optimized Driving Strategy
These three areas offered the most potential in allowing the team to surpass the efficiency performance of the 2013 vehicle. Defining these parameters allowed us to move forward as a team with a concrete set of design priorities.
I led the vehicle frame design process with a hyper-focus on mass. I aimed to create a frame that minimized mass while also providing the mechanical stiffness and stability necessary to ensure that our vehicle was both safe and dynamically stable. These priorities led to the frame design pictured below. We chose to use 6061 T6 aluminum, which offered a light weight solution that also provided sufficient structural properties and high machinabilty. We avoided the use of steel due to its increased mass density and low machinability. Machinability was key in allowing us to accommodate an intricate design and finely tune our mass optimization features. The main parallel frame rails were manufactured out of 1" x 2" hollow rectangular bar stock with an 1/8" wall thickness. The rectangular cross-section allowed us to concentrate mass far from the central axis of bending, which was critical in reducing mass and enabling the rails to withstand the loads they would encounter with a driver on-board. We were able to further optimize the mass of the rails by CNC machining a truss-like support structure along the sidewalls. This removed material from the sidewalls while maintaining sufficient load bearing material above and below the central axis of bending. Although they are not pictured below, tubular carbon fiber struts were added to support the aluminum roll bar - creating a triangular support structure that would further prevent buckling and better distribute loads during a roll event. We iterated and theoretically validated our design using SOLIDWORKS' CAD package and FEA tools. Final validation was completed through real world tests with our fully functional vehicle.
Designing the body of our vehicle provided the team with an opportunity to focus heavily on the aerodynamic efficiency of our vehicle. I acted as a Co-Project Leader within a group of six seniors who focused our final senior design course project around designing and manufacturing the body of our vehicle.
Our team utilized ideation sessions with specific design needs in mind to evaluate various design options. These design needs, along with a scoring matrix that was utilized to determine the form factor of our body design can be seen pictured on the right. The scoring matrix played a key role in making the decision to pursue a body design that left the front wheels of the vehicle exposed. Initial CFD analysis helped us conclude that this form factor would allow for a reduced cross-sectional area and drag coefficient, both of which are directly proportional to the drag force felt by the vehicle.
We went through further iterations of body designs to develop curve geometries that would best optimize aerodynamic performance. This required us to pay careful attention to flat surfaces that may create stagnation points, sharp features that may result in flow separations, and disproportional contours that may create pressure differentials across the vehicle's body. All of these factors increase the drag force felt by the vehicle and subsequently reduce its aerodynamic performance. Our initial iterations were created in CAD and analyzed using SOLIDWORKS' CFD analysis tools as shown to the right. We used the results of this analysis to select candidates for scaled wind tunnel testing with 3D printed models (see below). This allowed us to further validate our CFD results and move forward with an optimized body design that exhibited a 40% reduction in cruising speed drag force when compared to the previous year's competition vehicle.
With a body design defined, we moved on to the manufacturing stage. We continued our focus on mass reduction in choosing to build the body out of carbon fiber. This decision allowed us to create the complex geometries included in our design while utilizing a material that provided exceptional rigidity at an extremely low mass. The body was manufactured from scratch - starting with a hand-made male plug which was used to lay the fiberglass that would become our female mold and finally culminating in the hand-laying of carbon fiber to create our vehicle body. The pictures below illustrate the manufacturing process that we took on to bring our body design to life.
The Driving Strategy
The final parameter that our team strived to optimize was our driving strategy. I took this effort on as an individual project and led all aspects of the process. Before diving in, I created a methodical structure to guide my analysis:
Create a mathematical model of the DEMT's vehicle
Experimentally characterize the efficiency of the vehicle's motor
Hypothesize and theoretically test several driving strategies
Validate theoretical results using real world testing
I first built a simplified mathematical model of our vehicle using Matlab. This required the development of a time marching algorithm, which utilized the equation of motion shown on the right. The vehicle equation of motion incorporates several mechanical and dynamic properties of the vehicle, which were characterized using a combination of theoretical analyses and experimental tests. For example, the rolling friction of the vehicle was approximated using a ramp-based passive-rolling test illustrated below on the right.
After developing a mathematical model framework, I focused on characterizing the efficiency of the DC brushless motor used to power our vehicle. This effort led to the experimental prony brake pictured below, which allowed me to gather data that characterized the efficiency of our motor at varied torque and speed combinations. The resulting efficiency data can be seen in the plot to the right below.
I used the experimental efficiency data illustrated above to build an efficiency surface fit that was subsequently incorporated into my mathematical model. With our motor efficiency characterized, I was able to test throttling strategies and evaluate their effects on the theoretical efficiency of our vehicle. One important constraint to remember is that the Shell Eco-Marathon requires all vehicles to average at least 15 mph for a given run to qualify for official scoring. This constraint played a large role in outlining the potential strategies that we could employ. I specifically evaluated the projected efficiency of a constant speed strategy versus a "pulse-glide" strategy in which the driver periodically accelerates to a high speed and then allows the vehicle to coast down to a lower speed before quickly accelerating again. My theoretical trials clearly showed that the "pulse-glide" strategy enabled our vehicle to exhibit superior efficiency. More specifically, it demonstrated that high input torques during pulsing along with moderate to low amplitude pulses provided increasingly efficient results (see table below). These qualitative conclusions were further substantiated through real-world testing, during which the pulse-glide methodology produced a 12% efficiency gain over the constant speed strategy.
The sprawling effort outlined above forced myself and everyone involved to stay focused on the efficiency of our vehicle. Our structured design methods allowed us to execute on our design goals extremely well and enabled our team to earn 2nd place out of more that 30 prototype battery electric teams at the 2014 Shell Eco-marathon competition. Our vehicle recorded an efficiency of 310 mi/kWhr, a 38% increase over the previous year's vehicle.