Probabilistic based evaluation of the structural reliability achieved for a typical building designed according to SANS 517:2009 and SANS 10162- 2:2010

Oosthuizen, Frederik du Toit (2011-12)

Thesis (MScEng)--Stellenbosch University, 2011.

Thesis

ENGLISH ABSTRACT: This study aims to perform a quantitative probabilistic based evaluation of the reliability achieved in the design of Light Steel Frame Buildings (LSFB) when designed according to the loading code, SANS 517:2009 and the new design code for cold formed steel sections, SANS 10162-2:2010. The evaluation was done as follows: A specific structure, chosen and designed according to the specifications given in SANS 517:2009, was modelled in a structural analysis program. From the analyses done it was possible to identify the most critical element for given failure modes. Spread sheets according to SANS 10162-2:2010 were developed to calculate the resistance or design values for the different failure modes. By using a First Order Reliability Method (FORM), the reliability index for each failure mode could be calculated and evaluated in three different ways. Firstly, the reliability margin implied by the design load was evaluated. It was assumed that the resistance of the profile had a deterministic value while the loads applied to the structure were taken as probabilistic, i.e. following their known distribution functions. From this evaluation it was found that the necessary level of reliability was achieved for all failure modes. Secondly, the reliability margin implied by the resistance of the profile was evaluated. The resistance of the profile was taken as probabilistic with a distribution function that could be determined from the known distribution functions of the profile parameters responsible for the capacity of the profile. The loading was assumed to have a single deterministic value. From this evaluation it could be seen that a very low level of reliability was achieved for the failure modes of shear working in on the strong axis of the profile as well as interaction between bending and axial load. This is due to the strong dependence of this failure mode to the thickness of the profile, to which no partial factor is applied in the design process. Thirdly, the reliability margin implied by both the resistance and loads was evaluated. In a real life situation both loads and resistances would have variability. The resistance and loading values were taken as probabilistic with their known distribution functions. From this evaluation it was found that the necessary level of reliability was only achieved for shear working in on the weakaxis and axial load. All other failure modes achieved a level of reliability slightly lower than the target level of reliability for South Africa. The stiffening effect of wall cladding elements were not taken into account in the analysis. The reliability of connections was also not evaluated. It can be concluded that the element reliability achieved through the use of above-mentioned codes seems to be slightly less that desired. There could be an argument for recalibrating the partial factors to achieve the desired level of element reliability. However, the cladding elements provide significant additional stiffness to the structure and there is no immediate cause for stiffness concern. Future studies should aim to quantify the contribution that the cladding elements make to the overall structural reliability. The influence of connections reliability should also be investigated.

AFRIKAANSE OPSOMMING: Die studie poog om ‘n kwantitatiewe probabilisties-gebaseerde beoordeling van die betroubaarheidsindeks vir Ligte Staalraam Strukture (LSS) te bepaal wanneer dit ontwerp word volgens die belastingskode, SANS 517:2009 en die nuwe ontwerpskode vir koudgevormde staal profiele, SANS 10162-2:2010. Die beoordeling is as volg gedoen. ‘n Spesifieke strukturele model is gekies, ontwerp volgens die spesifiekasies in SANS 517:2009 en toe gemodelleer in ‘n struktuur analise program. Vanuit die analises was dit moontlik om die mees kritieke element te vind vir gegewe falings modusse. Sigblaaie, volgens SANS 10162-2:2010, is ontwikkel om die weerstand van die profile te bereken vir die falings modusse. Dit was moontlik om die betroubaarheidsindeks op drie verskillende maniere te bereken deur gebruik te maak van ‘n Eerste Orde Betroubaarheids Metode (EOBM). Eerstens is die betroubaarheids speling wat deur die belasting geimpliseer was, bepaal. Daar is aanvaar dat die weerstand van die profiel ‘n deterministiese waarde het terwyl die aangewende belasitng as probabilisties geneem is met hul bekende verdelingsfunksies. Uit hierdie beoordeling is gevind dat die nodige vlak van betroubaarheid bereik word vir alle falings modusse. Tweedens is die betroubaarheids speling wat deur die weerstand geimpliseer was, bepaal. Daar is aanvaar dat die weerstand van die profile ‘n probabilistiese verdeling het wat bepaal is uit bekende verdelingsfunksies van die profiel parameters verantwoordelik vir die kapasiteit van die profiel. Daar is aanvaar dat die belasting ‘n enkele deterministiese waarde het. Uit hierdie beoordeling is gevind dat daar ‘n baie lae vlak van betroubaarheid is vir skuif in die rigting van die sterk as, asook interaksie tussen aksiaal-las en momente. Dit is te danke aan die falings modus se sterk afhanklikheid van die dikte van die profiel. Daar word egter geen parsiële faktor aan die dikte toegeken in die ontwerp proses nie. Derdens is die betroubaarheids speling wat deur beide die weerstand en belasting geïmpliseer was, bepaal. In die werklikheid sal beide belasting en weerstand ‘n vlak van onsekerheid hê. Die weerstand en belasting is as probabilistiese waardes geneem met hul bekende verdelingsfunksies. Uit hierdie beoordeling is gevind dat slegs die falings modus vir skuif in die rigting van die swak as en aksiaal-las die nodige vlak van betroubaarheid bereik. Al die ander modusse het steeds ‘n redelike hoë vlak van betroubaarheid. Dit is egter steeds laer as wat voorgeskryf word vir Suid-Afrika. Die verstywings-effek van die bekleding is nie in hierdie ondersoek in ag geneem nie. Die betroubaarheid van die verbindings is ook nie bepaal nie. ‘n Gevolgtrekking kan dus gemaak word dat die element-betroubaarheid wat bereik word deur die bo-genoemde kodes effens laer is as die gewensde. ‘n Argument kan ontstaan vir die herkalibrasie van die parsiële faktore om die gewensde vlak van betroubaarheid te bereik, maar die bekleding bied ‘n noemenswaardige addisionele styfheid aan die struktuuur. Daar is dus geen onmiddellike kommer oor die styfheid van hierdie strukture nie. Verdere studies moet poog om die bydra van hierdie bekledingselemente tot die betroubaarheid van die struktuur te kwantifiseer. Die invloed van die konneksies tot die betroubaarheid van die struktuur sal ook ondersoek moet word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/17792
This item appears in the following collections: