Among civil engineering constructions, which are still usually designed on the basis of linear elastic computation of forces, but inevitably require nonlinear methods to evaluate safety in failure circumstances, are cooling tower shells. Due to their large scale their stability has always been scrutinized with utmost care, while in the last two decades it turned out that the material type of failure is usually much more probable. Limit load multipliers obtained from nonlinear analysis of progressive fracture and reinforcement plastification are much lower than the critical loading factors from stability determinants.
In the paper two similar smeared cracking models, implemented in finite element codes DIANA and ANKA, are used to simulate the behaviour of a real cooling tower shell that, due to a technological error during the erection, faced a catastrophe after some years of service. A 1m high batch of concrete filled into the moving formwork at the level of about 30m had the grade (compressive strength) of 11MPa, which resulted in a formation of two openings (25m and 14m long) with buckled reinforcement and a danger of total collapse. The shell was immediately repaired basing on engineering experience and linear elastic buckling analysis.
Now, in the nonlinear analysis under dead weight and wind load, the original (designed), damaged and repaired shells are analyzed. Only cracking of concrete and plastic flow of the reinforcing bars are considered, leaving the behaviour of concrete in compression elastic. Finite element discretization with eight-noded degenerated shell elements requires programming of a special pre-processor in order to efficiently generate the input to the codes. Simulations of progressive fracture are shown and compared, leading to a reassessment of the structure's safety and some observations about the used models and codes.