Ocena mehanskih lastnosti betona po izpostavljenosti povišanim temperaturam z uporabo različnih regresijskih modelov

Urška Dolinar (Author), Tomaž Hozjan (Author), Gregor Trtnik (Author)

Abstract

V članku predstavimo možnost ocene mehanskih lastnosti betona po izpostavljenosti povišanim temperaturam z uporabo različnih regresijskih modelov in rezultatov neporušnih preiskav. Iz nabora različnih neporušnih preiskav smo uporabili ultrazvočno (UZ) metodo in metodo določanja sklerometričnega indeksa, ki se lahko uporabita neposredno na armiranobetonski (AB) konstrukciji po požaru. S poznavanjem mehanskih lastnosti betona po požaru, določenih z omenjenima metodama, lahko kasneje ocenimo nosilnost preiskane konstrukcije. Rezultati temeljijo na lastni, obsežni eksperimentalni raziskavi, ki je zajemala izdelavo preizkušancev različnih betonskih mešanic. Te so se med seboj razlikovale po vodocementnem (v/c) razmerju, vrsti uporabljenega cementa ter količini dodanega superplastifikatorja. Preizkušanci so bili izpostavljeni različnim povišanim temperaturam, in sicer 200 °C, 400 °C, 600 °C in 800 °C. Po ohladitvi na sobno temperaturo smo na preizkušancih opravili različne neporušne in porušne preiskave. Eksplicitne zveze med rezultati neporušnih in porušnih preiskav, opravljenih na betonskih preizkušancih po izpostavljenosti povišanim temperaturam, smo izboljšali z uporabo umetnih nevronskih mrež. Izkaže se, da na ta način lahko podamo dobro oceno mehanskih lastnosti betona po izpostavljenosti povišanim temperaturam. Hkrati je ta ocena dovolj dobra že v primeru uporabe rezultatov UZ-metode. Oceno pa lahko izboljšamo z dodajanjem informacij o mešanici betona ali najvišji doseženi temperaturi.

Keywords

beton;povišane temperature;ultrazvočna metoda;metoda sklerometričnega indeksa;nevronske mreže;tlačna trdnost;

Data

Language:	Slovenian
Year of publishing:	2020
Typology:	1.01 - Original Scientific Article
Organization:	UL FGG - Faculty of Civil and Geodetic Engineering
UDC:	691.32:539.411
COBISS:	21295107
ISSN:	0017-2774
Parent publication:	Gradbeni vestnik
Views:	331
Downloads:	64
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Estimation of mechanical properties of concrete after exposure to high temperatures using different regression models
Secondary abstract:	The paper presents the estimation of predicting mechanical properties of concrete after exposure to elevated temperatures using different regression models and the results from non-destructive test techniques. From a set of different non-destructive techniques, the ultrasonic (US) and rebound number techniques were selected, as portthey can be used directly on a reinforced concrete (RC) structure after exposure to fire. Based on known mechanical properties of concrete after exposure to fire the residual load-bearing capacity of the structure can be estimated. The results are based on extensive experimental study that was carried out on concrete specimens of different concrete mixtures that differed in water to cement (w/c) ratio, the type of used cement, and the amount of added superplasticizer. Next, the specimens were exposed to various high temperatures, i.e. 200 °C, 400 °C, 600 °C, and 800 °C. After the specimens had cooled down to the ambient temperature, non-destructive test techniques were performed, followed by destructive ones. The explicit relationships between the results of non-destructive and destructive test techniques, performed on concrete specimens after exposure to high temperatures, were improved by using artificial neural network (ANN) approach. It was established that by using the ANN approach, good estimation of the mechanical properties of concrete after exposure to high temperatures can be made. At the same time, this estimation is good enough only if the results of the US method are used. However, the estimation can be improved by adding information about the concrete mixture or the maximum temperature reached.
Secondary keywords:	concrete;high temperatures;ultrasonic method;determination of rebound number;neural networks;compressive strength;
Type (COBISS):	Scientific work
Pages:	str. 152-162
Issue:	ǂLetn. ǂ69
Chronology:	jun. 2020
ID:	12046439