Novel Subsystem Prototype

The client came to PA Consulting to de-risk a concept for a complex subsystem that essentially needed to push a noodle through a garden hose. I worked with another engineer to mature the concept from just an idea to a fully automated proof of principle prototype that could simulate movements based on data recorded in clinical trials.

Noodle bending

We needed to understand how a semi-rigid noodle would behave as the system pushed upon it, and one of the main things to consider was the friction experienced by the noodle. Through empirical experiments, I soon found out that the noodle could curl into a helix within the volume of the support structure, contacting it in many points along its length. Pushing in this state can cause the device to dig deeper into the walls of the hose, making it impossible to advance.

A diagram showing different buckling modes of a cylinder constrained in a tube. The helical buckling mode can enter a self-locking state.

I ended up finding the solution in an unexpected place: oil rig drill strings. Oil rigs use steel pipes to transmit both drilling fluid and torque to the drill bit. A full drill string may have hundreds of pipe segments joined together, forming a kilometres long string that behaves not too differently than our noodle. Though the drill string papers I was reading dealt with kilonewtons and kilometres, I found that the analytical models could be adapted to the millimetre scale I was working at. I was able to determine the maximum push force that avoids helical buckling, and determine the other key parameters affecting the noodle’s behaviour.

Three roughnecks were needed to manipulate this drill string segment.

A diagram showing a drill rig setup. The bottom right shows helical buckling and locking of the drill string. Surprisingly analogous to the problem I was facing.

Straightening the hose

The hose has residual bending due to the way its stored. Storage conditions can also introduce creep, further exacerbating this effect. The hose can be straightened with force, but can we do it with the available actuators? I worked with suppliers to understand the materials available for medical grade extrusions, honing in on Pebax, a high-performance thermoplastic elastomer (TPE). I also explored composite structures to leverage the advantages of different materials.

Another factor was the way the hose was spooled. I initially explored helical spooling, which would take up less footprint than a spiral spool. However, it turned out that helical spooling introduces torsional stresses on the material that causes an additional residual deflection that needs to be corrected by force. These stresses can resolve as helical twists that increase in frequency when pulled on. I had to settle for a spiral spool design which had its own challenges.

A helically spooled hose gains more twists when stretched.

Enveloping the noodle

The system cannot function unless the noodle is fully surrounded by the hose, but due to the system architecture, the noodle must also exit the hose somewhere in the middle. A clever way was needed to allow the device to pass through the hose at a exit location fixed relative to the spool, but not fixed relative to the end of the hose.

I did several brainstorms and concept development sessions to identify potential solutions. I settled on a split hose, which could be split open at a specific location to allow the noodle to exit. This design required carefully selecting the hose material to be elastic enough to close itself after being split open. The noodle can potentially squeeze through any residual gap along the main length of the hose. This also meant any plastic deformation during splitting was unacceptable.

I explored more complex hose profiles, such as ones with wings that could be bent to reduce the splitting force. I also looked into composite extrusions, where materials with different properties are extruded together so we can leverage the unique advantages of each material. Because the extruded materials took 3+ weeks to procure, I had to balance analytical methods with rapid prototyping.

Test Bed

The work culminated in a robotic test bed that would prove the viability of the design. The test bed contained multiple force gauges to evaluate friction losses along the hose, and several actuators matching the expected degrees of freedom on the full system. I selected all sensors, all actuators, and developed all the controls electronics and firmware for the system. The test bed was capable of dynamically compensating for friction losses based on an analytical model I developed. I also built a software interface that allowed the test bed to read from CSV data recorded during clinical trials. This way the test bed could simulate being driven by a surgeon.

The test bed system block diagram. The blue blocks are firmware elements, the green are OTS electronics, and the yellow are electromechanical elements with sensors.

The subsystem performed well in the test bed, fully demonstrating all required functionality.