PSI - Issue 5

S. Sahnoun et al. / Procedia Structural Integrity 5 (2017) 1267–1274 A.Saifi et al/ StructuralIntegrity Procedia 00 (2017) 000 – 000

1268

2

A large number of industries and households require rapid non-destructive techniques for measuring or detecting scale deposited in water distribution systems: pipes, valves, pumps, boilers, etc. …. The metallurgical control technique provides a very accurate measurement of the thickness of the pipe. However, the overall procedure is very time-consuming and costly and destructive (Kang (210)). Apart from the preventive actions that can be taken to combat of scaling, such as the control of water mineralization, it is still not possible to have clean water without corrosive power. In the context of non-destructive testing of water pipes by infrared thermography. We will present in this work a simulation by the 3D finite element method, of the scale presence in a steel water pipe. We will study the effects of the geometric parameters of the pipe like the thickness and the diameter on the scale detection. After studying the detection, we will characterize the layer of scale by studying the effect of its thickness on the thermal response of the infected pipe. Then we will propose a model which allows estimating the scale thickness from the surface temperature of the controlled pipe. In the end, as a millimeter of limestone deposited in a pipe can cause an increase in annual energy consumption up to 9% (Trials carried out by the University of Illinois and the U.S. Bureau of Standards.), We will use the technique of artificial neural networks to increase the estimation precision of the scale thickness in the water distribution networks.

Nomenclature y

thickness of scale thickness of pipe diameter of pipe

e p

d

C

absolute thermal contrast

C m T def

maximum absolute thermal contrast temperature on the flawed zone temperature on the unflawed zone

T s

λ ρ

thermal conductivity

density

c

heat capacity heat flux density

Q

y i sim thickness corresponding to the i y i mod thickness calculated by proposed model n the number of points considered. x i the input of neuron j w ij weight of connection y j output of the neuron bj input bias p j potential of neuron R correlation coefficient d i the desired values s i the real output values

th point obtained by simulation

2. Estimation of limestone thickness in steel water pipes

2.1. Description of model We modeled the steel pipes by a cylinder of diameter d, of thickness e p and of length equal to l = 1m. On the inner face of the latter we deposited a circular cylindrical layer of the scale (figure 1).