(178e) Investigation of Pipe Cleaning Process Using Planar Laser Induced Fluorescence

Investigation of
pipe cleaning process using Planar
Laser Induced Fluorescence

Min
Zhang¹, Federico Alberini^1*, Kylee R. Goode¹, Katharina Roettger²,
Peter J. Fryer¹

   ¹School
of Chemical Engineering, University of Birmingham, Birmingham, B15 2TT, UK

  ²CPI, Wilton Centre, Wilton, Redcar TS10 4RF, UK

Removal of fouling is a big challenge in food
and fast moving consumer goods (FMCG) industries. Understanding of the removal
process will benefit product recovery and minimise the waste and energy required.
We have used Planar Laser Induced
Fluorescence (PLIF) to investigate the fluid mechanically driven
cleaning process of a 25.4 mm (i.e., 1 inch) diameter (D) Perspex
pipe. Configurations with a pipe length (L) of 420 or 780 mm were
investigated, giving a corresponding L/D ratio of 16.5 and 30.7,
respectively. Cabopol 940 suspensions of different concentrations with a
rheology behaviour fitting well to a Herschel-Bulkley model have been used to
mimic model food and FMCG products, and water has been used as a rinsing fluid.

The Herschel-Bulkley model is commonly
written as Equation 1 (Steffe, 1996).

                                         (1)

where τ is the shear stress, τ_y
is the yield stress,  is the shear rate, k
is the consistency index and n is the
flow index.

The yield tresses, τ_y,
of the suspensions used in this study ranged between 16 and 52 Pa. All
Carbopol suspensions were doped with Rhodamine 6G with a concentration of about
0.45 mg/L. Water was used as a rinsing fluid for removal of the Carbopol
suspension from filled pipework. Such configurations allowed illumination of
the cross-section of the pipe using a laser sheet perpendicular to the pipe
flow. Experiments were performed at water flushing rates ranging from 790 to
1895 L/h. The experiment was conducted by filling the pipe with a suspension,
then water was flushed through at a given volumetric flowrate. Meanwhile, a
cross-section of the fully filled pipe was illuminated using laser sheet and a
camera used to record the entire cleaning process. The system obtained 600-5000
images at a frequency of 25 Hz. Using the acquired image and
self-developed image analysis MATLAB script, detailed information of the
cleaning dynamics were acquired.

PLIF images show clear fluid breakthrough
pattern and different characteristics in removal of different products at pipe
cross-section, e.g., zero-yield material flows out easily with water,
whereas patch cleaning is clearly observed with yield materials. The cleaning
efficiency is quantified by (A_c/A₀)%, where
A_c is the area where the suspension has been removed/cleaned
up and A₀ is the initial pipe cross‑section area fully
filled with suspension. Cleaning time for each test is determined from the PLIF
images when cleaning reaches 90%. Experiment data are further analysed
following Palabiyik et al. (2018), by which they obtained a linear
relationship between the dimensionless cleaning time (ϴ) and a
ratio of material yield stress (τ_y) to wall shear stress
(τ_w). The dimensionless cleaning time and wall shear
stress could be estimated following Equation 2 and 3, respectively.

                            
          (2)

where t_c is cleaning time
obtained from experiment,  t_R is the mean water residence
time in the pipe, U is the water velocity, L is the pipe length
and D is the pipe inner diameter.

                                        (3)

where

                                       (4)

where ρ,
U and Re is the rinsing fluid density, velocity and Reynolds
number, respectively.

Plots of ϴ against τ_y/τ_w
for the two configurations studied are shown in Figure 1. As it can be seen, a
linear relationship is obtained for each configuration, however, the
correlations for the two configurations do not follow a same relationship, due
to changes with pipe length.

Figure
1 Dimensionless cleaning time (ϴ) against a ratio of yield stress to wall
shear stress (τ_y/τ_w) for a short pipe and a
long pipe of 420 and 780 mm, respectively. τ_y of Carbopol
suspension varied from 16 to 52 Pa. Water volumetric flowrate changed from 790
to 1895 L/h.

The results in Figure 1 suggest that the fluid
mechanically driven cleaning process investigated in current study is not a simple
mass transfer. The above observation extends the approach of Palabiyik et al.
(2018). Data analysis shows that the cleaning process can divided into stages: (i)
a big chunk of suspension is removed at the central part of the pipe, which
leaves a layer of the suspension attached to the inner pipe wall; (ii) the
layer at the interface with water is removed gradually, so the thickness
reduces with time; (iii) a thin layer of the leftover suspension peels
off the pipe walls.

Experiment results show that it is feasible
to use PLIF to record fluid breakthrough pattern at pipe cross-section and to monitor
the entire cleaning process. Data obtained after a detailed PLIF image analysis
can give some quantitative information on cleaning efficiency, cleaning time,
factors influencing the cleaning efficiency, etc. Therefore, PLIF will
be used in future work to gain a further understanding of the cleaning mechanism
in removal of different types of products from filled pipe.