Abstract:
The majority of building structures in Sri Lanka, are of reinforced concrete. Often it is
required to estimate structural member sizes at the initial stage of a building construction
project for load evaluation, cost estimation and reinforcement design.
For member dimension estimation, theoretical knowledge alone is inadequate as there are
some practical issues also to be addressed. This study proposes reinforced concrete member
sizes for future projects based on design data of past low to medium rise buildings in Sri
Lanka.
Standards available for member size estimation, difficulties encountered by the designers
while following available standards, gaps in existing sources, and applicability to local
conditions are discussed. Structural and architectural drawings of twenty one buildings from
two to thirteen storeys were used to extract design details related to slabs, staircases, beams,
columns and footings elements. Data gathered for each element type are used to interpret
relationships between member dimensions and design parameters.
Artificial Neural Networks (ANN) is an artificial intelligence technique for recognizing
patterns among data that are difficult to represent algorithmically. This study also explores
the potential of using Sri Lankan design data from past buildings in Artificial Neural
Network models for predicting reinforced concrete member sizes.
Recommended structural member sizes are presented in graphs and tables, and compared
with the ANN model results. Finally the member sizes recommended by the study are
compared with the sizes derived according to the available literature.