ShipConstructor Drives Automatic Welding Robots

It wasn’t too long ago that SSI and Wolf Robotics demonstrated some co-development which automatically drove a fully autonomous welding robot to weld several ship panels. SSI develops Autodesk based solutions for the shipbuilding and offshore industry including ShipConstructor software, an AutoCAD based CAD/CAM product line; and EnterprisePlatform, a tool for sharing product data model information. For its part, Wolf Robotics has been integrating robotic welding and cutting systems since 1978.

The collaboration was part of a National Shipbuilding Research Program (NSRP) Computer Aided Robotics Welding (CAR-W) project with U.S. shipyards. Prominent among those participating shipyards was Bollinger, who today is testing that cutting edge technology, hoping to be up and running by mid-year, as it builds the U.S. Coast Guard’s all-important Fast Response Cutters (FRC).

According to SSI’s R&D Manager Patrick David, the project focused on looking for ways to reduce manpower costs, as well as standardizing and producing better quality welds. He explains, “The big problem for shipbuilding in terms of using robotics is that there is a high degree of product variability. Even though ships of the same class are going to be very similar, the nitty-gritty differences are enough that it wreaks havoc on trying to set up a purely automated manufacturing system/assembly line.” And, with those obstacles in mind, SSI, Wolf Robotics and Bollinger Shipyards embarked on the remarkable project. And while this effort, to date, isn’t the end of the journey, it likely represents a quantum leap forward in shipbuilding efficiencies on this side of the big pond.

Bollinger Shipyards is a longtime and active participant in the NSRP. This particular project began when Dennis Fanguy, VP of Quality Management Systems at Bollinger Shipyards was contacted by Wolf Robotics in mid-2014 to see if Bollinger would be interested in leading the NSRP CAR-W project. “Bollinger agreed to lead this project and submitted the proposal in September of 2014 and we were then awarded a contract to proceed in June of 2015,” explained Fanguy, adding quickly, “The objective of this NSRP project was to close the automation gap between domestic and foreign shipyards by eliminating the core programming bottleneck currently preventing broad industry adoption of robotic welding automation.”

The team then set out to develop software algorithms for high impact weld types centered on robot reachability, collision avoidance and auto-path generation kinematics that use electronic CAD Model and Welding Process Planning data. A gap analysis of the required changes between Bollinger’s current processes and the generic industry CAR-W process map was developed by the team. Today, says Fanguy, “We are developing a roadmap of alignment for other shipyards to follow. Finally, the project will attempt to leverage and help develop an electronic infrastructure that captures critical process knowledge (in electronic data form) related to welding so that it can be electronically leveraged by designers for Design For Robotics (DFR) applications, as well as by process planners and engineers.” Unspoken in all of that, it is also true that process data capture and transfer of welder knowledge is a critical risk associated with the aging welding work-force and welder workforce shortage in the shipbuilding industry.

SSI eagerly agreed to participate in the project. That’s because, Pat David told Marine News, “First and foremost, what makes us unique in our own space is that we are an AutoCad based product. There are far more qualified people out there who could utilize this software. Secondly, our costs are much lower overall as a product – more functionality, more capability, delivered to a designer at a lower price point than any of the other options. We’re proactive, we have good relationships with the yards and a lot of us used to work in these yards with the people that are still there. We know what they need and why.”

Beyond this, says David, SSI tries to be as agnostic as is possible when it comes to data requirements. Dealing with data built in other software would be a small step for the robot to convert that data. He adds, “It’s not a huge issue.” 

But before anyone spent any money, NSRP asked SSI and its collaborative robotic team to clarify a couple of things. One of those things was what kind of return on investment would the shipyard see and how long would that take? David explains, “So, we went to Bollinger to investigate what they were doing with current weld processes so we could understand where we could realize the most savings for them. We asked: what kind of welds to you do, how big, how often, and what kind of manpower does it take to accomplish that? We determined the most common type of weld that would have the most impact against the lowest capital cost to implement for the project. We targeted those types of welds.”

Already up and running and producing real results in Bollinger’s, the workflow starts in ShipConstructor by automatically identifying the welds in the ShipConstructor 3D CAD model. The ShipConstructor user then uses an interactive 3D visual drawing to configure the weld properties such as the weld standard and includes weld breaks, weld pitch, etc. The weld information and various other relevant data (e.g. geometry of panel to be welded) is exported to Wolf Robotics via SSI’s EnterprisePlatform. The operator on the shop floor reads the files generated in Robot Studio which runs a path planning algorithm that evaluates several collision avoidance scenarios. 

The effort was an R&D project for SSI, says David, so the end benefit for his firm is probably a little further down the road than that which Bollinger is now enjoying. But that doesn’t mean that SSI isn’t happy with what they’ve helped to accomplish. They are. “The accomplishments here are very specific, perhaps not readily apparent to someone who doesn’t fully understand the work flow or process of shipbuilding,” said David, adding, “It boils down to ShipConstructor outputting the geometric data – the CAD model data that is created with our product, along with the welding information that is also created in our product. You’re looking at the assembly that has to be manufactured and you’re looking at all the 3D model parts that you’ve created in your CAD system. Along with just seeing the geometry that is there, we are also able to identify within ‘Shipcon’ where the welds would have to take place, as well as the type of welds that they should be. That information is brought into the path planner on the wolf side to plan all of the welds that the robot will perform. What’s significant here is that any variability on the production side can be repathed very quickly on the robot side and with little manual intervention and in a minimal amount of time. The path planning software is that robust. The information is married into one package, exported into the robot.”

Bollinger is the only shipyard testing – and using – this technology today. Bollinger bought the robots, got the overhead gantries into place and rigged all of the equipment. And says SSI’s David, the success of the project at Bollinger has raised enough eyebrows at other shipyards that the NSRP project’s next phase will involve more than just one robot.

According to Bollinger’s Fanguy, the project team generated ROM (rough order of magnitude) costs of the system concepts and worked with Bollinger to develop ROI and payback estimates using Bollinger’s plans for implementation to apply timing to estimated savings and utilizations. An adoption curve was applied to reflect the rate at which CAR-W systems could realistically ramp-up to expected production capacity. This adoption curve was meant to capture the timing for organizational alignment, information flow infrastructure, personnel training, etc. The adoption curve used shows 0 percent adoption for 2015 during development, 1 percent in 2016 with prototype systems, 20 percent in 2017 for CAR-W production applications, and 80 percent and 100 percent in 2018 and 2019 respectively.

Fanguy explained the numbers further, telling Marine News, “From a high-level, a $4.5 million investment in robotic automation per yard ($72 million industry-wide) over a three to seven year period (based on the development pace of Computer Aided Robotics algorithms for high-impact weld types) could elevate the industry to a point where 15 percent of the welding volume in an average shipyard (estimated at 60 miles) could be completed with robotics. This investment is projected to save the industry in excess of $80 million annually at a return on investment in excess of 100 percent once the robotic cell or gantry is operating at full capacity by the second year.”

To be fair, both return-on-investment and payback period metrics are expected to fluctuate by yard based on the pace at which robotic systems are brought to full production capacity. Additional financial savings not addressed in this business case include benefits derived from reduced welder ergonomic issues. Costs associated with over-welding, including additional man-hours, the increased ship weight, increased hours of material grinding in the case of rejected welds, and other variables were not considered in the calculation.

In the end, says Fanguy, the project team demonstrated core competency with key technologies and the capability to execute a complex, multidisciplinary development roadmap. He added, “Significant progress has been made in Phase 1 toward realization of a flexible, efficient, high-productivity robotic welding solution for the U.S. shipbuilding industry. Opportunities to transfer this technology to additional shipbuilding and, broader defense applications continue to emerge. It is incumbent upon the Navy, shipbuilding industry, and U.S.-based suppliers of these technologies to provide a solution that will continue to place the United States at the forefront of innovation and manufacturing capability.” 

Asked where the technology and robotic welding would provide the most value, Fanguy replied, “We believe that certainly it is better suited for series build, but the intention at Bollinger is to use this technology for part families that may not require a series build contract to exploit the savings from this technology. And, he adds, “Based on the high level of commonality between the part families and weld types among both Bollinger and other Shipyards, it appears that there are significant technology transfer opportunities for technologies outlined in this assessment report.”

Pat David also thinks that the robotic approach has legs. In fact, he insists, “One of the issues prior to this project when using robots for manufacturing that had high variability was the amount of time that it took to program the robot to do the weld. Our ability to quickly generate that path – that information – is where the savings lie. So those past problems aren’t as valid as they once were. A lot of folks are going to reconsider this possibility now. Automated welding is nothing new. But, the barrier to entry has been significantly lowered with and due to this project.”

David continues, “The biggest obstacle, hands down, is the upfront capital investment.” And he agrees with Fanguy, saying, “This isn’t just limited to series-build hulls – one-off projects can significantly benefit as well. What used to take many hours for the programmer or the robot to generate a path for the welding now is exponentially faster than it used to be.” That’s good news for boatbuilders. As NSRP continues to lead, and shipyards and other stakeholders collaborate, everyone wins. That reality has never been clearer.

(As published in the February 2017 edition of Marine News)