Euro-SiBRAM’2002 Prague, June 24 to 26, 2002, Czech Republic
Session 6
Storage tank corrosion, loading scenario and probability of failure
Prof. Ing. Bøetislav Teplý, CSc.
BUT Brno, ikova 17, 66237 Brno, Czech Republic
Abstract
To recall the quantified risk definition an example of steel storage tank is analysed. The probability of reaching not allowable stress level is assessed taking into account the corrosion process and two loading scenarios.
Key Words: steel corrosion, storage tank, probability of failure, risk quantification
The utilization of simulation techniques is (and will be) encountered frequently in the risk assessment. The general definition of quantified risk reads
R = pf . C (1)
where C is an average value of expected damage, which could happen with a probability pf (often called the probability of failure). C can be expressed as financial value, in number of dead or wounded and/or other damage indicator. This definition has been already used for many years – see e.g. [1]. The form (1) will have to be applied in many cases (mainly in cases when the risk is associated to an individual equipment or structure).
E.g. according to the European Seveso II Directive [96/082/EEC] and Czech Republic law No. 353/1999 it is required to work out a Safety report encompassing the risk analysis of a serious accident caused with dangerous chemical substances. The paper shows an examples of the risk assessment associated with the degradation process (corrosion) of cylindrical steel tank storing benzol.
A non-uniform wall thinning of vertical storage tank at the corrosion rate of vc happens due to a corrosion effect. This rate was observed to be different above and under the level of stored medium (access of oxygen). The original wall thickness s of a storage tank is reduced during time t in average to the thickness of
sf = s – vc . t [mm, mm/year, year] (2)
As the s and vc values are quantities of some dispersion, they should be considered as random variables.
To illustrate a risk analysis considering possible leakage of dangerous substances and a scenario of a loading history of a storage tank the following example is described:
A vertical storage tank with diameter D = 16 m, height h = 15 m is embedded on a solid foundation. It is made of five shell courses with 3 m in height each. Two lower courses wall thickness is s = 12 mm and of three upper courses the wall thickness is s = 10 mm. The circular flat bottom thickness is sb = 14 mm.
Designers assumed the filling of the storage tank to levels from 12 to 15 m (scenario A). For the purpose of this study another loading history was chosen - during period ta = 30 years, the tank was filled to the levels 9 m up to 11 m only, and later, up to the designed level (scenario B). In both cases the levels of filling were considered to be random in some extend. Corrosion mean value of the corrosion rate within the surface level region was 0.17 mm/year, at regions constantly under the level 0.045 mm/year. These rates were considered as random values too. The yield point Re of the material has been considered as random within the range from 200 MPa to 240 MPa.
As the main purpose of this study is to show the approach only, the elastic stress analysis has been used. The limit function has been accordingly formed as the difference of actual stress and Re/1.5 and the probability of failure than computed for different time steps using the crude Monte Carlo method (due to the simplicity of the limit function). In real situation a plasticity analysis would be adequate the rupture of material being considered as limiting point. It is believed that the results of both these risk analyses would be proportional as far as the comparison of scenario effects are concerned.
The table shows some results gained for scenario A at the bottom part of the tank (the maximal stress is dominant), and scenario B for liquid level (here the greater corrosion rate proved to be dominant). It is evident that at the bottom the danger is lower; at the liquid level scenario B it governs. The history of loading cannot be omitted.
|
Probability |
of failure |
Age Years |
Bottom Scenario A |
Liquid Level Scenario B |
20 |
0 |
0 |
30 |
0.0003 |
0.0115 |
35 |
0.0314 |
0.0721 |
40 |
0.1050 |
0.2700 |
60 |
0.0777 |
- |
To assess the risk also the consequences of possible leakage has to be determined. In the case studied, the consequence would follow from the type of the liquid stored, its amount, the height at which the defect is encountered and from the situation in the surrounding inhabited area.
Acknowledgements
This study has been partially supported by Grant Agency of Czech Republic, project No. 103/02/1161.
[1] Stanislav Vejvoda, Bøetislav Teplý: Mathematical description of material damage of chemical equipment as a part of risk assessment. FAILURES 2002: Proc. of the Fifth International symposium Risk, Economy and Safety, Failure Minimisation and Analysis, Umhlanga Rocks, South Africa 27-31 May, 2002, pp. 95-106.