Issue: Volume: 24 Issue: 3 (March 2001)

Flying deep

By Diana Phillips Mahoney

Hold on to your seats, we're going down." These are not words passengers want to hear from their pilot when in flight over the ocean. Yet the ominous phrase takes on a new meaning when it comes from the mouth of engineer Graham Hawkes, a trailblazer in the development of submersible vehicles for underwater viewing. Hawkes is the visionary behind the design of more than 70 percent of all piloted underwater vehicles ever built for research or industrial uses and more than 300 remotely operated vehicles (ROVs). In addition, one of his crafts, the Deep Rover submersible, holds the record for enabling the deepest solo ocean dive at 3000 feet.

The most recent incarnation of Hawkes' vision is a first-of-its-kind, untethered two-person submersible craft that will allow scientists to "fly" deep into the ocean and explore the domain in a way not possible using existing submersible technologies. Scheduled for completion early next year, the DeepFlight Aviator, as the new vehicle is called, will be able to maneuver with a speed and ease unmatched by existing submersible vehicles. Crucial to the development of the new sub has been the engineering team's reliance on a well integrated suite of automated design and analysis tools.

DeepFlight Aviator evolved from an earlier award-winning Hawkes design called DeepFlight I. Introduced in 1996, DeepFlight I was the first underwater craft to employ the lift principle of airplanes. The vehicle was designed with stubby, inverted wings that provide a "negative" lift to counteract the slightly buoyant characteristics of the sub and pull the craft down. At cruising speeds, DeepFlight I is neutrally buoyant and depends on its thrusters to drive it up or down. This principle, along with the vehicle's small size, allows the pilot to descend at a very fast rate to maximize bottom time.
The passenger compartments of the DeepFlight Aviator submersible vehicle were optimized for comfort. Structural analyses found areas of weakness, and the design was fortified accordingly.

In contrast, conventional submersibles and submarines reach depth slowly, typically using a negative buoyancy ballast system to control dive and accent. The ballasts are continually loaded until the craft is slightly heavier than the water it displaces. To return to the surface, the ballast is jettisoned. Often, using this mechanism, the submersible will spend 90 percent of its underwater time getting to and returning from the underwater research site.

The DeepFlight Aviator takes the airplane principle to a new level. It uses inverted airfoils and positive buoyancy to "fly" underwater with a freedom of movement approaching that of scuba diving. The vehicle offers the depth and underwater-viewing capabilities of a conventional submersible (minus the noisy mechanisms and bright lights that cause those underwater organisms capable of fleeing to do so) and the mobility of a submarine. Unlike conventional submersibles, which require a specially equipped mother ship to accommodate their size and crew, DeepFlight Aviator is designed to operate independently under on-board pilot control.

To function in such a manner, the vehicle is self-propelled by eight 24-volt batteries. In order for the airfoils to overcome the small sub's positive buoyancy, the vehicle must move at an underwater speed of eight knots (versus the two knot-limit rarely exceeded by conventional submersibles). Herein lies the core problem that Hawkes' engineering team had to solve in order to generate a functional design: how to optimize the trade-off between power, weight, and mass to enable the vehicle to quickly reach its depth and move freely once it has done so.
Structural analyses of the DeepFlight Aviator design helped engineers create a model with the integrity to withstand a water pressure load approaching 700 pounds per square inch at a maximum depth of 1500 feet.

Although water is frictionless, its extreme density means that huge power increases are required in order to effect incremental speed increases. For example, increasing an underwater craft's speed to five knots from one knot requires a 100-fold power increase. This need had to be balanced with the low power-to-mass ratio of batteries, which dictates a small, hydrodynamic shape to minimize drag. Along with these factors, the ability to endure pressures approaching 700 pounds per square inch (psi) that the craft would be subject to in reaching a target depth of 1500 feet, as well as passenger safety and comfort, had to be taken into account. To meet the unique demands of the project, the Hawkes' DeepFlight Aviator engineering team went digital-hook, line and, sinker.

In addition to being the first vehicle of its kind for the world of underwater exploration, DeepFlight Aviator is a first for Hawkes' company, Hawkes Ocean Technologies (HOT), in that the entire structure was designed and analyzed digitally in three dimensions. Previous HOT crafts have been built using 2D and 3D automated tools for some aspects of the design, but never before has the first and only prototype been virtual. In fact, the "finished" vehicle still exists only virtually, while various individual components are in different stages of the physical manufacture process.

The vehicle structure was modeled using Autodesk Inventor (tm), and the potential performance of each component was then analyzed separately and in concert using the finite-element analysis capabilities of Ansys DesignSpace to assure optimal efficiency, effectiveness, safety, and comfort.
DeepFlight Aviator, a first-of-its-kind two-person submersible vehicle, is designed to fly to its depth using positive buoyancy and inverted airfoils.

Conceptually, the craft can be likened to a rigid diving suit that employs the practical functionality and maneuverability of an ROV. Visually, it looks like a small, elongated spaceship with two viewing bubbles. The bubbles are actually individual pressure holes, one each for the vehicle pilot and passenger.

These passenger pods are the most critical component of the vehicle, and thus were subject to the most stringent design. "Obviously, since those are the human-occupied spaces, we can't have those things failing," says Eric Hobson, a mechanical engineer on the project. "The pods went through more than 100 major revisions where we reanalyzed them maybe 20 or 30 times each."

The reason for such scrutiny, says Hobson, is that the design deliberately avoided traditional shapes. "Typically, when you're building something that's going to be withstanding pressure, you want nice rings, cylinders, and spheres-shapes that spread the stress out evenly." In fact, he notes, "that's how DeepFlight I was designed, with a traditional cylinder head and bullet nose on both sides. The shape was optimized to make the most out of the least amount of material."

Unfortunately, a vehicle optimized for material can be downright uncomfortable. "[In DeepFlight I] the pilot lies down prone with his or her head first, and it can be difficult to maintain that position for any significant amount of time without getting muscle cramps," says Hobson. "We wanted to avoid that with DeepFlight Aviator because it is designed specifically as a passenger craft, with spots for the pilot and a second person. So comfort is really at a premium."

Consequently, instead of optimizing the pod shape for material, the engineers optimized it for the human body. "We went off the beaten path to get a shape that really forms around the human sitting position," says Hobson. "But whenever you do that, you really open the door to making a mistake, making something that won't be able to withstand the pressure."
The pressure load on each of the passenger pods was calculated based on the maximum potential depth of the vehicle during use, then applied to the digital model to assess its performance.

To avoid such disaster, the engineers used FEA to structurally retrofit their chosen cocoon-like pod design, which consists of a cylinder that tapers from 26 inches in diameter at shoulder level to 18 inches at the feet. "For the pod, we essentially cut a cylinder at an angle, to get a hoop that puts stresses equally around the whole ring. Then we cut that in half and ended up with this thing that just wants to collapse on you because it's not totally supported," says Hobson. The structurally weak model was analyzed to assess the stress concentrations, and material was added and shifted where needed to meet the structural demands without compromising the ergonomic considerations.

Each significant change to the model necessitated a new analysis to assess the effect of the revision and the need for additional changes. "It came down to a matter of massaging the shape to get all of the stress concentrations out of it," Hobson says. For example, the highest concentration of stress from the water pressure was calculated to be at the most inward curving section of the pod (the upper end of the pod cylinder is revolved 30 degrees to accommodate the seated pilot). Based on these results, the engineers designed the cast-aluminum pod to be three inches thick at that point, versus one-inch thick near waist level and three-quarters of an inch thick by the feet.

Fortunately, changes made to the Inventor model could be brought directly into DesignSpace. "That's why we chose these tools, because they're so well integrated," says Hobson. "We were able to seamlessly attach our models to the DesignSpace database and retain all of the associativity when we changed the Inventor models. Without this tight associativity, the constraints, loads, and supports would have to be reapplied to the geometry of the model in the analysis software after almost every change. This becomes hugely time-consuming for the critical, complex designs, when we might be running 20 to 30 iterations."
Through associative links between the DeepFlight Aviator model and the analysis program, engineers were able to tweak the design to meet optimal safety conditions.

The ability to assess each of the components both individually and within the assembly context was also essential to the design project's success, says Hobson. "With a system this intricate, the results from analyses of individual components would be meaningless." For example, while it was important to determine that the metal locking ring that clamps the fittings for the acrylic cover to the passenger cavities would not deform under 670 pounds of pressure, that information says nothing about the seal for system integrity. "That's very important," says Hobson, "and there's no way to know that without analyzing the entire pod with all of its parts." Ultimately, the craft's deep-sea worthiness and the FEA results will be put to the test physically at a US Navy facility.

Once the engineers were satisfied with the overall design of the vehicle and the performance of the pod and its constituent parts, they made the unusual decision to cast and machine the pod first, before the designs of the remainder of the vehicle had been finalized. "We roughly made the sub, all the components, the frame, the wings, everything, but just in terms of the outer shape. When we were happy with that, we started fully designing the components individually," says Hobson.

The rationale behind this approach, he notes, is to rein in CAD's potential for exploring limitless design possibilities-a prospect that can easily consume all of the "productivity enhancements" that make the technology appealing. "With today's advanced tools, it's easier than ever to revise and reform a design, but at some point, the value added is negligible. There's always the possibility that there might be a slightly better configuration, regardless of how many design iterations you go through," says Hobson. It becomes more efficient, he says, to settle on a design that meets your needs and push it through, even though doing so reduces overall design flexibility, since a significant change to any of the remaining parts would likely alter the stress distribution across the entire vehicle and thus necessitate a change to the pod.

"You do limit yourself by working this way, but sometimes that's a good thing," says Hobson. Otherwise, he adds, designers risk getting mired in the process, losing their focus on the end product.

When the vehicle is finally assembled, it must meet American Bureau of Shipping certification standards, after which it will be ready for its first adventure: a school for underwater aviation, where, says Hobson, participants will learn to pilot the craft. "We want to get people excited about the underwater world, and the best way to do this is by letting them see it first-hand."

Diana Phillips Mahoney is chief technology editor of Computer Graphics World.