Knowledge Base: Don’t Forget the Basics…
Capabilities of modern pre-processors
Mark Chillery, Chalice Engineering, UK

Article provided by NAFEMS, www.nafems.org

Types of Pre-processor
Earlier or ‘traditional’ FEA pre-processors used a simple approach to building geometry wherein points are defined, which are connected by lines, which are, in turn, used to define surfaces. If solids are required, these were defined from a previously defined closed set of surfaces. More modern pre-processors emulate 3D CAD functionality and build 3D models by starting out with a basic 2D sketch on a face or construction plane and then carry out operations such as sweeping, lofting, reflecting, cutting etc.

In some traditional pre-processors, if a point is used in the definition of a line, then moving the point may not be permitted without deleting the line. If there are several lines meeting at the point and these in turn are used to define surfaces, simply moving a point can quickly become prohibitively difficult. In contrast, if the point can be moved without deleting any higher level geometry and their definitions are updated to reflect the new point position, then a considerable time saving will result.

In the traditional pre-processors, sweeps, revolves and other operations can be carried out, just as for CAD emulating systems. But a crucial difference between the traditional and CAD based approaches is that in the earlier pre-processors, the underlying geometry is hierarchical and can only be modified by changing the point definitions. In the case of a sweep command for example, if the resulting geometry was not correct, it is likely that the best approach would be to delete all new geometry (or use an undo command) and recreate the sweep command. In a 3D CAD based package, however, access to the part history means that a sweep operation can be modified and retried, without recourse to basic geometry modification and the likely reworking this entails.

In principle, complex geometry is possible via traditional pre-processors, using ‘sweep’ and ‘revolve’ and ‘cut’ commands. However, these systems can sometimes run into tolerancing problems and databases become corrupted. Any analyst should be able to manage such issues by regular saving and backups. Ideally though, prospective purchasers should trial a pre-processor on models similar to the intended use, to determine the capabilities of the package being assessed.

In light of the above, if using a traditional package for analysing a relatively complex 3D structure, the analyst has to make a critical decision whether to model a reduced complexity version from scratch within the preprocessor, or import the geometry and somehow deal with the likely numerous small features that prevent an efficient mesh from being created. Even modern CAD based pre-processors will not be able to import all CAD formats.

Importing the geometry can be quick to try. Sometimes, import attempts may not even produce a satisfactory duplicate of the model that existed in CAD in the first place. The use of FEA graphical applications that can import native CAD based modelling formats like ‘Parasolid’ will have an advantage here, over those that only import for example, IGES or other non-native formats. Efforts to resolve such geometry import issues are often a major project schedule uncertainty in any situation where the link between the CAD software and FEA pre-processor is untried.

Using a CAD emulating pre-processor usually means that design modifications can be much quicker than in the traditional package as, for any given geometry modification, more of the original solid and FE model definition is retained so mesh and load updates are often instant or require little rework before a new run can be commenced.

Integrated Analysis
A third alternative to the above types of pre-processor is carrying out the analysis within the analysis functionality integrated into a CAD modeller. Such packages tend to focus on ease of use above analysis visibility. The current state of the art is dominated by static analysis, although multiple components can be modelled effectively as a fused single body. Features such as meshing and restraint application with little user direction are common. ‘Chunky’ solid components are best suited to an integrated analysis but for models that are best suited to plate or shell analysis, it is often not appropriate to represent these with solids. In other cases, the analysis can become too large.

Such analyses can be a useful design tool for quick checks although the ease of use of such tools prevents extensive interrogation and hence confidence in the results is often low. Very often in practice, such analyses are not acted upon, unless attempts are made to incorporate them more formally into the design cycle.

Geometry Import Issues
Data management
If transfer of CAD data from a CAD system to an analysis system using intermediate file format such as IGES, STEP, ACIS or 'Parasolid' is carried out, the administrative connection between the CAD and analysis is lost and there are then questions over data management of the disconnected analysis data and result files. Previous designs may be required to be reworked and it is necessary that storage and naming of files are retained at least in the short-term. Very often, all previous analyses are forgotten until it is time to write a report. Unless some provision for recording past work is adopted, it may be difficult to demonstrate the validity of the current final design by referencing results from previous analyses in such a report.

Very often analysis process and associated files fall outside any existing data management systems, In this case, an analyst should consider their own pre-emptive file naming system, so that the relevance of files can be understood later.

Feature suppression
Many features present in a CAD model are not required in the ensuing FEA and can be ignored or suppressed. Fillet radii, holes and anything that can be considered a stress raiser in areas that may have high stresses cannot be omitted. But if this is known to be a non-critical region, all such features can be removed. The easiest approach with a traditional system is to suppress these features in the CAD system and regenerate a simplified model for analysis. This would be preferable to creating new geometry to fill or smooth these features in the pre-processor.

Virtual Topology
As well as the presence of unwanted real features, there is also the issue of small surfaces or lines existing in most CAD models. While these are usually unnoticeable in the CAD environment, they can force the meshing process to create correspondingly small element edges. This can cause distorted and overly dense meshes, or sometimes cause the mesh to fail altogether. Most pre-processors have tools for highlighting or plotting all small surfaces or short lines (for example, by finding all lines below a certain factor of average line length). These are useful because they are often impossible to detect by inspection of the whole model. Whilst most ‘traditional’ pre-processors have this facility in some form, they often have few effective tools for eliminating them. In extreme cases short edges and slither surfaces can only be eliminated by redefining the volume by creating new surfaces. For even a moderately complex solid model, reworking the geometry in this way can be a tricky and sometimes time-consuming process.

As a result of the above issues, the increasingly common tools for automatically eliminating such unwanted small features from the model are available in many packages and are the key to successful import and meshing. Often termed ‘virtual topology’, these tools permit satisfactory meshing operations which might have been impossible without their use.

Contact:
Mark Chillery,
Chalice Engineering.
mark.chillery@chalice-engineering.com

Design Plateau Key to A350 Goals
Robert Wall, article provided by Aviation Week & Space Technology
 
By year-end, Airbus plans to nail down the central supplier, industrial and design decisions for its A350XWB twin widebody. Not since the A300B, the first aircraft the European aircraft maker brought to market, has so much been riding on getting it right.

With the A350, Airbus is trying to overhaul its design, program management, and supply chain processes. The goal is to cure itself of ills that have bedeviled the aircraft maker in the past, ranging from configuration conflicts within different parts of the company to repeated late deliveries on recent products that have marred its reputation with customers.

The general consensus is that through paying heed to specific input from under-whelmed customers, the A350 has evolved from a nonstarter into a major force in the marketplace. Indeed, customer apathy to the original A350 is credited with making the European manufacturer realize it had to raise the bar to counter the 787.

Much of what Airbus is trying to do is riding on a new design approach, using a virtual engineering plateau to manage all aspects of the program. The move to the multi-disciplinary tool has already yielded long-term savings in development and production time, says Francois Caudron, vice president for A350 customer and business development.

And it’s worked. Emirates CEO Tim Clark says he’s reviewed the plateau concept and is optimistic it will allow Airbus to deliver the A350 as promised, despite the aggressive schedule the aircraft maker has laid out for the project.

The concept, which involves nearly 850 employees, includes working on a single digital mockup (DMU), updated nightly, to ensure configuration control. The use of different DMUs and design tools on the A380 caused configuration problems that significantly delayed assembly; this time there is standardized CATIA V5 software across company disciplines, from engineering, production, program management and finance to customer communications.

There are at least two clear examples where the plateau idea has paid dividends, Caudron points out. First, as engineers were sorting out how to split the fuselage sections, production representatives pointed out that shifting the joints away from heavily loaded areas would yield big savings in final assembly. Under the serial approach, that ramification may not have been noticed for months.

Similarly, the process of optimizing the A350 nose involved multiple parties. While aerodynamicists were primarily focused on the shape, materials experts helped define the metal and composite configuration mix, and systems experts ensured there was enough room for six large displays. The commercial side then made sure the crew rest area would not detract from the revenue space of the cabin. All these decisions, formerly made in a serial fashion, are now made in parallel.

There are other signs the new program management approach is working, Caudron notes. Airbus, by the end of 2007, had managed to select 70% of the system suppliers as planned, even though bidders were skeptical that this would be accomplished. By the end of March the level is anticipated to be at 90%.

While the A350, like the A300, may have much riding on it for Airbus, there is a big difference. The A300 was slow to attract customers; the A350 has sold strongly and its order book is at well over 300 aircraft, illustrating it is competitive against main rival Boeing’s 787.

Airbus Chief Commercial Officer John Leahy contends that the A350 is the fastest selling airliner, although the program began with an inflated baseline of around 100 orders for an earlier design iteration of the A350.

Interestingly, several customers have opted for both the 787 and the A350. While in some cases the decision has clear political overtones—for instance, when Aeroflot bought some of each product both manufacturers made large industrial commitments to Russian industry—with others, such as Singapore Airlines, that political impetus doesn’t apply.

Alan Pardoe, A350 marketing director, suggests a market split may be evolving between the 787 and A350, with the airlines that are looking to replace 767-300s focusing on trip costs and leaning toward the Boeing product, while those focused on seat-mile cost are tending to favor the A350. It’s not a view shared by Seattle, where officials see no distinct split. In fact, Qatar Airways, also a purchaser of both, says the determining factors for them were simply slot availability and fleet renewal needs.

Airbus has customers to thank for more than just buying the aircraft; without their persistence, the aircraft may never have evolved into a competitive product. The aircraft maker’s initial response to the 787 was effectively to do nothing. Then, when the original A350 was unveiled, with an A330/A340-fuselage diameter cross section, many potential buyers pressed the aircraft maker to do more. Even with the subsequent, larger XWB configuration, there was some customer dissatisfaction.

This push for new design iterations resulted in a more costly development program, which management resisted. About eight months ago, EADS CEO Louis Gallois rejected an idea by International Lease Finance Corp. (ILFC) Chairman Steven Udvar-Hazy to install composite fuselage panels on a composite substructure, rather than an aluminum one. A few months later, Airbus adopted the configuration, noting that it had maintenance advantages. ILFC has since ordered 20 A350XWBs.

The shift to the XWB also gave Airbus the chance to put in place a new approach to building the A350, divorced from the A330/340 set-up. It revolves around using fewer suppliers with greater systems responsibility.

Honeywell was the first to be named; the company had been tapped to provide the auxiliary power unit (APU) on the original A350, but now will deliver the entire APU/air management system, which includes environmental control, cabin pressure, bleed air and supplemental cooling systems, along with starter generators and the APU installation kit. For Airbus, it means one point of interface, rather than dealing with many partners. It also offloads financial risk.

Moog will also take a central role, and several more of these types of relationships are expected this year.

Other program partners include Rolls-Royce, for TrentXWB engines; Goodrich, nacelles; Liebherr, nose landing gear; Messier-Dowty, main landing gear; and Rockwell Collins for trimmable horizontal stabilization actuators.

The partners are also part of the A350 design plateau, although Airbus has to firewall them to some extent due to competitive implications. For example, Honeywell will need to access the air systems areas without gaining insight into sensitive avionics data that may still be being competed.

Yet to be worked out are the exact allocation of major structural work packages that will go to the buyers of the six facilities EADS is unloading. Latecoere is taking over the sites in France, OHB Technologies is the new owner of the plants in Germany, and Filton will now hold the site in the U.K.

And a resolution is needed for some of the international partnering arrangements. Russia and China have each been told they’ll get 5% of workshare, while South Korean industry is scheduled to receive 3%. Exactly what elements of the airliner they will build still needs to be set. Airbus also wants to involve Japanese industry, hoping to take advantage of the large composites expertise there. Japan has also been a market long dominated by Boeing, so the A350 could represent a chance to break in.

Risk-sharing structural partners have already been involved in the plateau as part of the joint development phase. The winners will remain, losers will be dropped.

Even when the plateau is phased out, some of the features will remain. For instance, Airbus will maintain a single DMU to ensure all partners are working to the same baseline.

There already have been plenty of opportunities for the A350XWB schedule to fall behind, not least because naming the buyers of the Airbus sites has taken longer than planned. But Airbus management, all along, has been keen to insulate the A350 from such turmoil. A case in point, long-lead items, such as autoclaves, have already been ordered, even as the future of the facilities where they will be installed is pending.