Validation criteria for Sargon: Validation criteria for Sargon.pdf

Validation schedules of Sargon (excerpt): main_beams

Validation schedules of Sargon (excerpt): SARGON vs NASTRAN

Validation schedules of Sargon (excerpt): EN 1993-1-1 (Eurocode 3)

The increasing complexity of the software techniques available and of the code requirements, has made more and more complex and subtle model creation for the structural analysis (generally a finite element model) and its right use and management, actually its validation. 

On the other hand designers often do not have a deep knowledge of  computational mechanics (and especially of finite element method fundamentals) and numerical analysis, and have little or no knoledge of programming, and so a poor understanding of the typical problems related to the management of large amount of data.

The error search in a structural model is therefore a typical specialistic issue, requiring special know how and knowledges to be carried on.  The idea that computations made thanks to just one model and just one program can be considered, by definition, reliable due to the fame of the program used or due to the size of the Firm which made up the model, has no scientific base and therefore must be left out. Excellent programs can be misused and also big companies have flaws.

 Castalia has wide knowledges in this field and a wide experience (books, papers, case histories, some of which also published), which has revealed more than one time modelling problems and also programming problems in some softwares also widely used.  Castalia not only developes its own software tools (all widely documented with works  fully proving the high software reliability, in the limits in which this is today possible) but also owns the licence of very important high level software tools like NeNASTRAN©, NXNASTRAN© and FEMAP© (interfaced with  Sargon©) and moreover of average software tools like ALGOR©. We are therefore able to cross check systematically different solvers results, thus arriving to a real control of every single run, of every single model.

The ability of writing ad hoc software, tailor made on the basis of each model or test need, lets us execute checks and reliability validations that no end user, with no commercial program, coluld ever be able to perform.

All this could lead to just nothing if there was not a specific code trend aimed to made thirdy party checking of computations and models as compulsory. In Eurocode 0 (EN 1990:2002), appendix B, is said that for structures belonging to  "consequence class CC3" ("grandstands, public buildings where consequences of failure are high"), is  recommended as minimum requirement for the computation check  (table B4) the "thirdy party checking", that is the check "executed by "an organisation different from that which has prepared the design".

Castalia proposed since a long time that this kind of checks, the unique way to properly validate a result, is made compulsory for the most important designs. Only the cross checks of results got with different solvers and models can give a real idea of the effectiveness and reliability of the results got in cases which are not theoretically manageable.

Experts in finite elements and automatic checks are very few. In fact not only it is necessary to know deeply the matter (finite elements in linear and non linear field, codes, science and technique of constructions), but also having a wide programming experience and being able to develop specific software tools, interfaced with existing large programs, and able to perform some of the checks needed. Castalia has developed one of these tool, named checksolvers, since years: it is mainly used internally but could easily be used to check models developed with different solvers.

Error checking is a specific work requiring a very high expertise, and a know how that must embrace many fields: it is not enough being expert in structural engineeering or in programming, you must be expert in both, and must have developed a wide experience in this specific field. Castalia is one of the (few) Italian companies which has really the features to propose itself for these activities:

• having directly developed structural software (and not reselled programs made by others, which is pretty much differerent), always very rigorously and always without advertising its softwares in a misleading way;
• having published in a clear way books referring to and results got with its softwares, and comparative tests and benchmarks;
• having a wide experience in this specific field (civil, industrial, mechanical and naval fields);
• being at first place in deeply examining code texts, never accepted acritically, but always carefully examined and criticized in a constructive way (steel structures, seismicity).

The basis of these checks are the proposals made by Castalia in  "Structural software", a work in Manuale dell'Ingegnere 84^ edizione (so called"Colombo"), 84^ edizione, Hoepli, 2003 Castalia offers its computational tools and its wide know how to execute checks and validations of every model and/or computational tool, thanks to a specific quality testing service.

Milan, may 2007.