In the US, the shipbuilding industry has been shrinking due to the reduction in defence budgets. To counter these trends, the industry must increase its competitiveness and be capable of delivering high quality products under compressed schedules at very low cost.
To support these needs, the design systems at these shipyards must rely on their Information Technology systems to tightly integrate their design, business and manufacturing processes so that they can quickly make transitions from design to build with a known level of risk associated with costs and schedule.
Newport News Shipbuilding (NNS) (Newport News, VA), is the only shipyard capable of building Nimitz-class aircraft carriers and one of two capable of building the Navy’s nuclear powered submarines.
In February 1996, NNS launched a programme to reduce cycle times in processes it uses to design and build ships. Known as `Full Speed Ahead,’ this programme focused on various processes such as design, production planning, material sourcing, steel fabrication, and outfitting. Process innovation at the shipyard sought to go beyond incremental improvement and dramatically cut the time required to produce ships.
NNS uses a variety of CAD/CAM/CAE packages for many aspects of ship design but engineers perform all of their ship product modelling using the company’s proprietary, VIVID system. VIVID allows users to do concurrent design work in multidisciplined environments. Using the product modelling system, users design the vessel’s structure that is then used as a `background’ on which to place and arrange equipment, piping, heating/ventilation/air conditioning (HVAC), as well as electrical trunks and cableways.
When development began over ten years ago, VIVID was used to design ship structures using constructive solid geometry (CSG) primitives on applications developed and run on mainframe computers. According to Mike Polini, supervisor of software development, `The applications were suitable for visualising the majority of the ship’s structure but were deficient with regards to dealing with non-analytic surfaces such as the hull and for supporting the automated manufacturing processes required by robotic cutting and welding work cells.’
Needless to say, when NNS developed an object-oriented, UNIX-based VIVID, its focus was on improved structural design capability. Later, when robots were purchased to automate its manufacturing work cells, this required engineering data to be exchanged between designers and the robots. Workstation VIVID was designed to achieve that goal, as well as manage manufacturing for plate and profile parts.
`We realised,’ says Poloni, `we needed a geometric modeller to address surface definitions, surface-to-surface intersections, curve and surface intersections.’
The result was the selection of ACIS as its modelling kernel. By integrating ACIS into its application, it was hoped to gain a mature geometric modelling engine with a minimum of software development effort. `We are not a traditional ACIS user,’ Polini explains. `We do not store solid models in our database. We store stripped down representations that are more or less curves and surfaces with some attributes and features. Workstation VIVID is feature based. As the design is progressively refined, localised part modifications are made by adding features. Each part can derive a solid model of itself, as can each feature. These solid models are created on demand. The process of generating a final solid model of the part is accomplished by doing a Boolean subtraction of each feature from the part.
`At the manufacturing stage of the part’s life, weld shrinkage becomes a percentage feature on the part. We can adjust a curve or surface by scaling it based on a specified weld shrinkage value. We can determine, for instance, if the part requires a cut-out in a corner, how much material will be removed by the cut, and how the assembled parts will fit relative to one another,’ Polini adds.
Using VIVID, engineers store attributes on the underlying surfaces and curves representing the plate and profile parts. ACIS can then derive solid models using the attributes when needed. `Not every engineering or manufacturing process requires a solid model,’ reports Polini. `We can send data directly to the robot at the work cell because the robot has macros that recognise parameter values. We do not have to create a geometric representation to cut the part with the robot. Therefore, we don’t degrade performance and clutter up our database with solid models that we don’t need.’
Newport News’ definition of a product model includes information about the factory. VIVID allows users to route objects through various work cells that make up the factory.
Polini’s team established relationships within the product model between assemblies of parts for manufacturing and a workcell, such as a cutting robot, a welding robot, or an entire robotic assembly line, that will actually do the work. It is a relationship recorded in the product model. The work cells can ask questions of the objects assigned to them to find out if they are appropriate for processing at this site. This capability allows NNS to proactively address potential problems before they occur.
Polini says, `Of key importance to us is that ACIS serves as our curve and surface modelling engine. Because of the large number of parts – up to one million – necessary to define a ship model of an aircraft carrier, it is impossible to represent each part as a complete solid model. VIVID has been developed to represent these parts in a more compact form based on attributed curves and surfaces with features. Each part and feature is capable of deriving a solid model representation of itself upon demand.
`However, these solid models are not stored in the databases. This technique provides performance improvements over conventional mechanical CAD systems.’
ACIS-enabled Workstation VIVID was developed early in 1996 and was implemented in late 1996 to support the construction of a series of new commercial ships.
`When we began building commercial ships,’ Polini says, `we committed to installing the robots in the factory. The only way we could use the robots effectively was if we implemented VIVID to deliver data to the robots. When the design of the ship was started, we began developing the software system. We loaded the definition of the parts and activities and fed the information to the robots. They cut the parts and put the ship together.
`We adjusted the software to reflect changes in how ships would be designed by factoring in new capabilities such as robotic cutting and welding. We were doing a lot of re-engineering, software development, machine installation, and ACIS implementation simultaneously,’ says Polini.
INFORMATION: Spatial Technology Tel: +1 303 544 2900