Sei sulla pagina 1di 1

132

J.M.F.G. Holst and J.M. Rotter

and this is only important if it becomes very large. Such tilting is discussed by Greenwood (1974) and Bell and Iwakiri (1980). Their recommended limit is given as 0.5% (i.e. a tilt of 1:200), although Greenwood (1974) suggested that larger tilts could be accepted without causing binding of a tank onto a oating roof. The effect of such a tilt is that the pressure is reduced on one side of the tank and increased on the other, leading to global bending, but not to ovalisation of the tank. However, near the uid surface the pressure becomes zero around part of the circumference, which does result in ovalisation. The above conclusion is therefore sensitive to the aspect ratio of the tank, and the problem is more serious in very low height to diameter ratio tanks. The components of settlement remaining after the uniform settlement and the uniform tilt have been removed are usually smaller, but are more important. These can only be determined with reasonable precision if a rather large number of settlement observation stations are set up around the tank (to be useful, the absolute minimum number is eight, but sixteen stations would be preferable, as noted below). These settlement components are referred to here as differential warping settlements. When examining differential warping settlements, it is important to establish rst whether the tank has a xed roof, a oating roof or no roof. For tanks with a oating roof or no roof, it is necessary to consider the stiffness of the wind girder or ring at the eaves. Where the tank has a xed roof, the roof plays a major role in the response of the tank to the vertical displacements beneath the wall (differential warping settlements). A xed roof must be included in any structural model of the tank. Characterisation of components of wall settlements In their eld study, Sullivan and Nowicki (1974) concluded that tanks with diameters up to 110 m could accommodate differential warping settlements as large as 30 mm without problems, but that values in excess of 45 mm caused problems irrespective of the settlement distribution or tank diameter. Such a nding appears rather sweeping and takes no account of the form of the tank or the effect of differential settlement on it. A simple crude method of assessing the severity of local differential settlements is to take the difference between the settlements at any two adjacent measurement stations around the tank perimeter, and divide it by the circumferential separation between them. This is sometimes termed the angular distortion. A value of 0.35% was suggested by De Beer (1969) and Belloni et al. (1974). The latter authors improved on this and used the maximum change in slope between three adjacent settlement observation points under the wall. They recommended that this slope be limited to 0.22% as a working hypothesis for satisfactory operation of large tanks with oating roofs. Unfortunately, the procedure leads to quite variable results depending on how many observation stations are used. The criterion is also not rationally founded.

Potrebbero piacerti anche