FUOYE Journal of Engineering and Technology, Volume 3, Issue 1, March 2018 ISSN: 2579-0625 (Online), 2579-0617 (Paper) FUOYEJET © 2018 12 http://dx.doi.org/10.46792/fuoyejet.v3i1.146 engineering.fuoye.edu.ng/journal Stability Analysis of Fluid-Conveying Beams using Artificial Intelligence *Theddeus T. Akano and Olumuyiwa S. Asaolu Department of Systems Engineering, University of Lagos, Akoka, Nigeria {takano|oasaolu}@unilag.edu.ng Abstract— This paper employs artificial intelligence in predicting the stability of pipes conveying fluid. Field data was collected for different pipe structures and usage. Adaptive Neuro-Fuzzy Inference System (ANFIS) model is implemented to predict the stability of the pipe using the fundamental natural frequency at different flow velocities as the index of stability. Results reveal that the neuro-fuzzy model compares relatively well with the conventional finite element method. It was also established that a pipe conveying fluid is most stable when the pipe is clamped at both ends but least stable when it is a cantilever. Keywords— ANFIS, fatigue, fundamental frequency, natural frequency, stability —————————— ◆ —————————— 1 INTRODUCTION ipes are of great importance in both domestic and industrial fields including oil and gas (especially fluid conveying pipeline as in Figure 1), aerospace, nuclear, industrial processing, as well as power generation and transmission. They are used for conveying various fluids from one point to another. The transmission of fluid in pipes results in induced vibrations due to the inherent properties of the fluid. Fluid-induced vibrations resulting from deflection experienced on the walls of the pipes causes high instability and fatigue of the pipes. This instability can lead to events which are not suitable for the state of life and property in nuclear reactors, mines, oil pipelines, heat exchangers and other pipe using industries. Studies have shown that pipes become prone to instability and fatigue when the first fundamental natural frequency becomes zero (Baohui et. al., 2012).The effects of the system parameters on pipe behaviour were studied by (Modarres-Sadeghi & Païdoussis, 2009). Clearly, the natural frequency of a pipe decreases with increasing velocity of fluid flow (Grant, 2010). The need of properly understanding causes of instability to pipes conveying fluids is of great importance. These instabilities sometimes cost so much to the general system. The natural frequency of a pipe reduces as flow velocity increases; this is because as flow velocity increases, the pipe stiffness reduce, causing a drop in the fundamental natural frequency of the fluid-conveying beam. In general, the entire stability of the pipe conveying fluid depends on the natural frequency of the pipe and its critical velocity (Chellapilla & Simha, 2008). * Corresponding Author The deployment of Artificial Intelligence into many fields has brought endless possibilities to solving problems. Previous researchers have shown how it has been used to predict, monitor and control vibration problems especially as experienced in pipes conveying fluids. Zahra et al. (2014) used the artificial neural network to monitor vibration of Steam Turbine in a Nuclear Power Plant. A lot of research work has aimed at understanding the mechanics of pipes conveying fluid. The concept for modelling the interaction between fluid and structures from aero and hydro-elasticity has shown the applicability of computational methods for problems involving flow-induced vibration (Zilian, 2014). Besides, experimental studies have shown the correlation between the fluid flow in a pipe and the resulting vibrations. Investigations have shown that pipe vibration levels are proportional to the fluid flow rate (Safari & Tavassoli, 2011). Computational analysis was used to determine the critical fluid velocity that induces the threshold of pipe instability when conveying fluid (Grant, 2010; Seo et al.(2005); Ritto et al.(2014)). Farshidianfar, (2012) described the instability experienced in pipes as a result of the fundamental natural frequency vanishing. In their own works, Hakim and Abdul Razak, (2011), and Kao and Hung (2005) employed Artificial Neural Networks (ANNs) in the dynamic stability analysis of damaged and undamaged structures. The present study employed ANFIS in predicting the stability of a given pipe when conveying a given fluid at a given fluid flow speed. Different boundary condition scenarios were also looked at. 2 MATHEMATICAL AND INTELLIGENT MODELS 2.1 Equation of Motion for Fluid-Conveying Beam Mathematical models for the dynamics of a straight fluid-conveying beam are developed. These models were used to derive the fundamental natural frequency, critical fluid velocity and natural frequency. P Fig. 1: Alaska oil pipeline (Source: google.com)