Analysis of Acoustic Fatigue in Safety Relief Systems

To protect installations in the gas and oil industry and process industries against overpressure the installation of relief systems is common practice. However, there have been incidents where such systems fail as a result of high dynamic loads caused by: the generation of shock waves when the relief valve is opened very instantly; the occurrence of two-phase flow in the relief line and the corresponding formation of slugs; instability of the relief valve and acoustic fatigue. This paper focuses on acoustic fatigue, which occasionally has been the cause of incidents like at Krechba – In Salah Gas Project Algeria where a breakdown of a relief system occurred. The extremely high sound levels caused by the expanding flow through a relief line finally caused fatigue failure of the system. Such incidents are the reason for the requirement to analyze the possibility of acoustic fatigue of relief systems and other systems with a strong acoustic source already during the design. In the paper we present the method that TNO applied for acoustic fatigue analysis. So far a calculation of the internal sound power levels was made and evaluated against the allowable levels presented by Carucci and Mueller [1]. However, this method does not allow computing the cyclic stresses of the piping. Due to the large dimensions and the high frequencies involved, it is not possible to compute accurately the sound field in the piping and to use a deterministic method to evaluate the cyclic stresses of the piping. In this paper we present work recently done where the Statistical Energy Analysis (SEA) method is applied to evaluate the cyclic stresses in the piping material. Due to the statistical nature of the method, it is not possible to determine with deterministic certainty the stress levels in every point of the piping. Yet, this method gives a range of cyclic stress for each pipe section or element. This leads to a more reliable criterion for acoustic fatigue studies than the earlier method, which is based on maximum allowable pressure fluctuations.