Aging natural gas pipeline infrastructure is becoming an increasingly large problem in the United States. There are more than 2.4 million miles of pipelines currently in use, all of which require regular maintenance and inspection to ensure safety. It is estimated that 70% of these lines were installed prior to the widespread use of the most common inspection tool, pigs, and therefore require some other tool to carry out tasks such as direct line inspection, pipeline mapping, gas quality monitoring, and cleaning. This has prompted a large growth in the area of robotic inspection devices to fill this market gap. However, many of the robots developed either fall short of true autonomy, are unable to operate in live flow conditions, or are designed for only a specific pipe size. This thesis details the design of a robotic platform called MARPI, or Miniature Autonomous Robot for Pipeline Inspection, which addresses the weaknesses of both pigs and previous robots. MARPI is a wheeled robot that was developed to include several key features: energy harvesting, wireless communication, onboard navigation system, and a small profile and footprint in the pipe. The robot uses two 150:1 micro gear motors for its drive mechanism and features a permanent Neodymium magnet to make the robot adhere to the surface of steel pipes. The energy harvesting system was characterized through a series of wind tunnel experiments which showed that to maximize the power generated it is best to have a turbine with a high number of buckets/blades, streamlined bucket geometry, and a relatively large offset from a bluff body below. To carry out the design of MARPI, a statics model was developed and used to predict the magnetic force required to adhere to and avoid sliding in the pipe, and the motor torque required to propel the robot. This model was used to analyze the performance of the robot as a function of robot size. Key results show that to minimize power consumption, the robot should travel vertically with the flow, and to maximize range per day, a small robot with a large turbine is best.
