Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Detection of fouling in food pipelines using Long Range Ultrasonic Testing (September 2009)

   
Phil Catton, Septimonette Chan, Vichaar Dimlaye, Peter Mudge

TWI

Anke Fendler, James Turner, Gary Tucker
Campden BRI

Paper presented at NDT 2009, Blackpool, UK, 15-17 September 2009.

Abstract

The suitability of Long Range Ultrasonic Testing (LRUT) to detect fouling in food and drinks pipelines has been tested. Fouling (the accumulation of unwanted materials on solid surfaces) is a major problem encountered in the food industry. Monitoring of deposit formation and scheduling cleaning procedures accordingly can improve efficiency and reduce spending by upgrading plant maintenance from a time based to a condition based schedule. Ultrasonic guided waves have demonstrated excellent sensitivity to coatings on surfaces of piping, and several authors have described the application of ultrasonics for fouling detection in the food industry. [1-7]

The deposition of chocolate and palm oil fats on cooled surfaces and the detection of these deposits by LRUT have been studied. Deposition experiments with chocolate were performed by melting chocolate onto a cool stainless steel pipe. Palm oil fat deposition experiments were performed in a circulating system, where melted palm fat flowed through cooled stainless steel pipe.

Ultrasonic guided waves were excited in the specimen and recorded with varying levels of fouling deposits present. Localised fouling was found to cause signal reflections, and signal attenuation was experienced with deposits spread more uniformly along the pipe, showing the potential for direct measurement of fouling severity.

1. Introduction

The present work aims to identify the NDE needs of the Food and Drink production industry, specifically in the area of product-carrying pipes. In consultation with food and drink production companies, a key point of interest is the potential of using Ultrasonic Guided Waves (UGWs) to detect and/or monitor fouling of pipe.

In the course of food production, particularly in processes involving heat, product can build up on the pipe surface, reducing thermal efficiencies and flow rates, and introducing the potential for burnt-on deposits to break off and end up in the finished article. This is clearly undesirable, and so susceptible pipes are periodically flushed through with cleaning chemicals, at intervals and for durations known from past experience to be sufficient.

As there is no real direct measurement currently employed, the cleaning process is not optimized. Not only are there costs associated with the cleaning chemicals themselves (and their disposal), but during cleaning the pipe can not be used productively. Minimizing the down time associated with cleaning therefore has clear economic advantages. Measuring the level of fouling can help achieve this in two ways. First, monitoring the build-up of deposits in the pipe may help to increase the interval between cleans (as cleaning may currently be conducted sooner than necessary). Secondly, monitoring the cleaning process may help reduce the cleaning duration (as the cleaning process may currently continue beyond the point when pipes are clean).

The development of a method to directly measure the level of fouling deposits in a pipe is therefore an attractive area of research. This paper details the experimental findings of initial development work.

2. Overview of ultrasonic guided waves

Ultrasound propagating through solid media has proven very useful in non-destructive testing, where waves can be excited in such a manner that they interact with geometrical or materialistic characteristics of a specimen. Detecting and analysing these interactions can provide useful information about the size and location of flaws or defects in the specimen, and so Ultrasonic Testing (UT) is used in a wide range of engineering industries. Tiny flaws are normally the subject of this type of inspection, and so UT usually operates in the Megahertz frequency range to provide the necessarily small wavelengths of the propagating sound. This offers very high resolution and sensitivity, however the distance over which waves of such high frequency can be expected to propagate is very small - of the order of centimetres. Therefore, UT is normally applied to specimens where detailed knowledge of a small area is required, for example the quality of a weld on a pressure vessel.

It is also possible, however, to inspect large components of an appropriate geometry using UGWs. Where structures exhibit an elongated shape, such as in the case of a pipe, sound can be excited in such a manner that it is constrained to only propagate along the length of the body. While still ultrasonic, they are excited at a very much lower frequency (usually between 20 and 200kHz) than conventional UT, such that the wavelengths are of a similar order of magnitude to the geometry of the test specimen. Ultrasound correctly excited in such objects can propagate many hundreds of meters and is reflected by changes in cross sectional area (such as girth welds or volumetric defects). UGWs have been used to screen many tens of meters for the presence of corrosion and other volumetric defects in petrochemical pipeline for several years - this is known as Long Range Ultrasonic Testing (LRUT). [8]

Despite the useful characteristic of UGWs propagating long distances (thereby allowing long lengths of pipe to be rapidly inspected from a single test position), the propagation distance is greatly reduced when an attenuative coating is present. Thick, heavy, and viscoelastic coatings affect the waves the most. It is for this reason that UGWs have the potential to be used to successfully detect and monitor the presence of fouling deposits in food process lines. The more heavily fouled a pipe becomes, the more the propagation of guided waves will be affected.

3. The technical challenge

LRUT is widely employed for rapid corrosion screening in the Petrochemical, and Oil and Gas industries, using commercially available hardware such as the Teletest Focus, GUL Wavemaker or the MsSR 3030 systems. The differences between piping used in these industries and the food industry are many, some of which are detailed in Table 1.

Perhaps the toughest challenge in porting LRUT to food and drink pipes is the pipe-in-pipe configuration. Heat exchangers are constructed such that an outer pipe has one or more smaller pipes within its bore. Product will usually travel in the inner pipes, with a heating or cooling fluid such as steam or water being passed between the inner and outer pipes. As the critical areas are those relating to the product itself, it is the inner piping that is of most interest, to which there is little or no access. This creates a practical conundrum of exciting guided waves in the inner pipe.

Table 1. Piping used in the oil and gas industry compared with that of the food and drink industry

 Oil and gasFood and drink
1 Usually carbon steel pipes Usually stainless steel pipes
2 High pressure products require pipe wall >5mm Pipe walls <1.5mm
3 Pipe diameters usually >100mm Pipe diameters usually <50mm
4 Pipe sections welded together Flanged pipe sections bolted together
5 Usually single skinned piping Usually 'pipe-in-pipe'
6 Problems due to fouling build-up uncommon Must be kept clean

As far as the propagation of UGWs is concerned, the geometrical factors (rows 2 and 3 on Table 1) have the largest effect. As radius decreases, so does the number of possible ultrasonic guided wave modes that can exist at a given frequency. In the frequency range of 20-100kHz, many tens of wave modes are possible in oil and gas pipes, but as few as 3 may exist in the smaller pipes used in food and drink production. Of these, a single wave mode can be selectively excited by appropriate array design. When so few wave modes are present, this can often be achieved simply by adjusting the plane on which transducer displacements oscillate. For example, the fundamental torsional wave mode T(0,1) can be generated to the exclusion of other modes by a single shear mode transducer displacing with a circumferential motion with appropriate pipe geometries and frequencies.

As such, the potential for practical problems is somewhat reduced by the ability to use a single transducer instead of the large, bulky and complex arrays required for more conventional LRUT of oil and gas lines. The solution proposed here therefore, is to incorporate a single transducer element at each end of a heat exchanger section, which can be installed at the manufacturing stage and permanently mounted to the inner pipe. Electrical connection can then be made via a simple gland arrangement. The results presented here assume this configuration, such that ultrasound is transmitted at one end of the pipe and received at the other end, as shown in Figure 1.

Fig.1. Experimental set-up to monitor fouling build up. A transducer was placed at each end of the pipe with a cooled area in the middle to encourage localised fouling
Fig.1. Experimental set-up to monitor fouling build up. A transducer was placed at each end of the pipe with a cooled area in the middle to encourage localised fouling

Clearly, a very wide range of products are manufactured in the food and drink industry, and each will cause a different type of fouling mechanism, which range from soft fatty deposits, to harder burnt on residues and scales. This paper documents both preliminary studies conducted to prove the concept of fouling detection using guided waves as well as more in depth trials. Chocolate was used for proof of concept, and palm oil for the detailed trials.

4. Investigations

4.1 Chocolate

Chocolate is not a product whose fouling is typically problematic in the food industry. However, it is widely available, solid at room temperature and melted with minimal effort, and cheap - and therefore an ideal substance to easily model a fouling system. As such, A new, 6m length of 101.6mm OD x 1.5mm wall 316L stainless pipe was used to simulate the pipes identified in food manufacturing facilities. Note that this diameter is much larger than might be expected in reality, although the wall thickness is similar. It is not thought that the diameter will have an adverse affect on the results of the present tests. A larger pipe was used here so that existing transducer arrays designed for the petrochemical industry (where there are typically much larger pipes) could be used with minimal modifications required. A transducer array was placed at the very end of the pipe and used in pulse-echo mode.

As these tests are intended as quick 'proof of principle' investigations, for convenience chocolate was added to the pipe in units of 0.23kg (ie, one bar at a time). The material was added as uniformly as was practicable, covering the entire circumference of the outside of the pipe over an axial distance of 0.55m at approximately one third along the pipe from the transmission end. Note that past experience has indicated that the ultrasound will behave the same regardless of if the fouling substrate is on the inside or outside of the pipe.

With the pipe set up in this manner, ultrasound was allowed to propagate along the length of the pipe, interact with any fouling substrate, and reflect from the far end of the pipe. A range of test frequencies were used from 20-100kHz to establish any frequency dependencies of the interaction with fouling.

Figure 2 shows two notable features that indicate guided waves can be used to detect and monitor for the presence of fouling on food pipeline. Firstly, it can be seen that the repeating echoes from the signal reflected from the far end of the pipe are attenuated very much more when there is a large amount of chocolate (fouling) present. Secondly, a small signal can seen corresponding to the location of the chocolate, indicating that the fouling deposit is not only attenuating sound, but also causing a small amount to be reflected back. Note that this is unlikely to be seen in real situations, as fouling tends to build-up uniformly in the pipe system, albeit perhaps with some areas more heavily fouled than others. Having shown the feasibility of such methods, investigations could then be planned for other fouling processes.

Fig.2. Frequency response from pulse-echo array for clean pipe (left) and pipe with 0.23kg of chocolate fouling (right). Attenuated repeating end reflections and reflection caused by the chocolate can be observed
Fig.2. Frequency response from pulse-echo array for clean pipe (left) and pipe with 0.23kg of chocolate fouling (right). Attenuated repeating end reflections and reflection caused by the chocolate can be observed

4.2 Palm Oil

A new, 6m length of 50mm outer diameter 1mm wall 316L stainless pipe was used to simulate the pipes identified in food manufacturing facilities. This was connected in a circuit through which palm oil was pumped at a low speed. In order to keep the fat fluid, it was heated to about 40°C. However, at this temperature, the oil flowed freely and caused only negligible levels fouling. To accelerate the build-up of fat deposits locally between the sensors (while still flowing freely through the rest of the circuit), the pipe itself was surrounded by ice at -18°C (as shown in Figure 1). As there is no easy method to directly measure the actual thickness of the fouling layer, a procedure was conceived whereby the liquid palm oil was periodically allowed to drain completely from the test pipe. This allowed it to be removed from the rest of the circuit and be placed on a sensitive set of weighing scales, so as to measure the total mass of fouling deposits within the pipe. Every 60 minutes the pipe was weighed in such a manner to within 0.1g. Signal measurements were taken every 2 minutes, where tests were conducted from 20 to 100kHz.

Example time history results for a clean and fouled pipe are shown as waterfall plots in Figures 3 and 4 respectively. From these figures, it is clear to see a significant difference in the distribution of received amplitudes over a range of frequencies, between the 'before' and 'after' cases. It can also be seen that there are a number of wave modes in operation, arriving at different times for different frequencies. Interestingly, the faster moving wave modes (ie, those arriving sooner on the time base) appear to increase in (relative) amplitude with increasing levels of fouling deposits. The slower moving modes appeared to behave somewhat independently of the amount of fouling. At a test frequency of 90kHz, a strong correlation between the amplitude of the faster moving modes and the amount of fouling was observed. This is illustrated in Figure 5, were a linear trend with a R2 value of 97% can be seen.

Fig.3. Waterfall plot of a range of frequencies for a clean pipe. At 1400µs, the arrival time of T(0,1) can be seen at all frequencies. There are slower propagating modes with high amplitude arriving after this. The signal from 100-500µs is an electronic artefact
Fig.3. Waterfall plot of a range of frequencies for a clean pipe. At 1400µs, the arrival time of T(0,1) can be seen at all frequencies. There are slower propagating modes with high amplitude arriving after this. The signal from 100-500µs is an electronic artefact
Fig.4. Waterfall plot of a range of frequencies for a pipe containing 2.8kg of fouling deposits. At 1400µs, the arrival time of T(0,1) can be seen at all frequencies. There are faster propagating modes with high amplitude arriving before this at some frequencies, but the slow modes have been attenuated. The signal from 100-500µs is an electronic artefact
Fig.4. Waterfall plot of a range of frequencies for a pipe containing 2.8kg of fouling deposits. At 1400µs, the arrival time of T(0,1) can be seen at all frequencies. There are faster propagating modes with high amplitude arriving before this at some frequencies, but the slow modes have been attenuated. The signal from 100-500µs is an electronic artefact
 Fig.5. Trend of relative amplitude of faster moving wave modes with weight of fouling deposited. Note a very good linear trend can be observed with an R2 of 97%
Fig.5. Trend of relative amplitude of faster moving wave modes with weight of fouling deposited. Note a very good linear trend can be observed with an R2 of 97%

5. Conclusions

Initial pilot empirical studies have determined that hard fouling in the form of chocolate can be easily detected using ultrasonic guided waves. This lead to the design of more technically accurate experiments using a through transmission method of testing. Heated palm oil was pumped through a cooled pipe and the weight of fouling deposited was periodically measured. Received signal amplitudes correlated very will with the weight of fat deposited, with a linear trend exhibiting an R2 of 97%. This implies that this method can be used to directly assess the amount of deposition build-up in food process lines, which could allow for a dynamic and reactive approach to pipe cleaning.

6. Future work

While high quality and robust experiments have been conducted with fatty deposits, only rudimentary preliminary investigations have taken place with harder fouling materials. In the coming months, further work will be undertaken using a similar experimental design, but heating the pipe while whey protein is pumped through. This should create a burnt on residue, measurement of which would be desirable. Similarly, scale deposition will be investigated by pumping hard water through the pipe, to model hard substrates.

7. Acknowledgements

Funding from the Technology Strategy Board is gratefully acknowledged (Aug 07 - Jul 10), in collaboration with TWI Ltd, Campden BRI, Cadbury UK, HJ Heinz Ltd, Coors Brewers Ltd and Plant Integrity Ltd. This project is lead by TWI Ltd.

8. References

  1. Withers, P.M.: Ultrasonics for the detection of fouling. Food Technology International Europe 1994: 96-99 (Reprint 768).
  2. Jo da Silva, J.; Wanzeller, M.G.; de Almeida Farias, P.; da Rocha Neto, J.S.: Development of Circuits for Excitation and Reception in Ultrasonic Transducers for Generation of Guided Waves in Hollow Cylinders for Fouling Detection.Instrumentation and Measurement Technology Conference, 2005. IMTC 2005. Proceedings of the IEEE Volume 1, Issue , 16-19 May 2005 Page(s):467 - 471.
  3. da Silva, Jaidilson Jo, Lima, Antonio Marcus Nogueira, da Rocha Neto, Jose Sergio: Fouling detection using ultrasonic guided waves and parameter estimation. Third IFAC Symposium on System Structure and Control (2007).
  4. Hay, T R; Rose, J L: Fouling detection in the food industry using ultrasonic guided waves. Food-Control, 14(7): 481-488, 2003.
  5. Lohr, K R; Rose, J L: Ultrasonic guided wave and acoustic impact methods for pipe fouling detection. Journal-of-Food-Engineering, 56(4): 315-324, 2003.
  6. McClements, D.J.: Advances in the application of ultrasound in food analysis and processing. Trends in Food Science and Technology, 6, 293-299, 1995.
  7. Withers, P.M.: Ultrasonic, acoustic and optical techniques for the non-invasive detection of fouling in food processing equipment. Trends in Food Science and Technology, 7, 293-298, 1996.
  8. Mudge, P, Lank, A, & Alleyne, D. 1996 (26-28 November). A long range method of detection of corrosion under insulation in process pipework.

For more information please email:


contactus@twi.co.uk