f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/fbp

Cleaning validation of fermentation tanks Satu Salo a , Alan Friis a , Gun Wirtanen b,∗ a b

BioCentrum-DTU, Kgs. Lyngby, Denmark VTT, P.O. Box 1000 (Tietotie 2), FI-02044 VTT (Espoo), Finland

a r t i c l e

i n f o

a b s t r a c t

Article history:

Reliable test methods for checking cleanliness are needed to evaluate and validate the clean-

Received 28 December 2005

ing process of fermentation tanks. The size of a fermentation tank can be several hundred

Accepted 18 January 2007

cubic meters limiting the use of many traditional sampling methods since the reachable areas are scarce compared to the entire tank surface. Pilot scale tanks were used to test the applicability of various methods for evaluation of the cleanliness of fermentation tanks.

Keywords:

The aims of this study were to find appropriate methods to detect cleanliness in pilot scale

Sampling

tanks, to find out which parts of the tank were difficult to clean and to optimise the cleaning

Tank

procedure. This study was performed to improve the investigation of cleanability of process

Cleanliness

scale tanks and to choose suitable methods and sampling areas for use in process scale evaluations. Furthermore interpretation of data from various sampling methods is needed in cleanability assessment. The methods found to be suitable for validation of the cleanliness were visual observation of a fluorescent indicator using UV-light, a contact agar method and culturing based on swabbing and swiping with non-woven cloths. The validation of the cleaning procedure and the design of a proper cleaning system can be supported and improved using computational fluid dynamics, which through a simulation reveals areas not easily covered by the cleaning fluid. Microbial results from pilot scale studies are needed to interpret the simulation results properly. Using various methods simultaneously improves the interpretation of cleanability of tanks. © 2008 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers.

1.

Introduction

The hygiene of production surfaces at dairies, breweries and beverage plants has a crucial effect on the quality of final products. Therefore, the hygienic requirements and the efficiency of the cleaning procedure as well as the frequency of cleaning and disinfection actions must be integrated with the process design. Hence reliable validation methods must be available to obtain relevant information concerning the hygienic state of processes plants. The traditional methods available all apply physical testing at the surface in question. However, the accessible surface area may be limited compared to the entire product contact surface. Therefore validation of the cleaning procedure is potentially problematic in closed processes, since most of the interior in big tanks are unreachable. The sizes of process tanks can be enormous leading to practical problems in cleaning as well as in physical sam-

∗

pling, i.e. it is impracticable to take a representative set of samples in tanks with a volume measured in hundreds of cubic meters. Also access to surfaces in pipelines, valves and pumps for biofilm residue samples is limited. According to statistical guidelines the number of samples needed for the evaluation depends on the amount of units (volume, area, pieces, etc.) in studied object, occurrence of searched factor, e.g. a pathogenic microbe, and reliability level of the research method (Anon., 2003). The statistical likelihood of presence of unacceptable products can be decreased by increasing the amount of samples tested. As an example of reliability of results according to statistical approach analysing and obtaining acceptable results from 5 samples out of a total lot of 100 it can be said with 95% certainty that not more than 45 pieces out of those 100 are unacceptable with respect to the analysed hygienic state (Anon., 2003). A practical problem can be that there are not enough sampling valves in long pipelines in a process line. Besides that the sampling valves

Corresponding author. Tel.: +358 20 722 5222; fax: +358 20 722 7071. E-mail address: gun.wirtanen@vtt.fi (G. Wirtanen). 0960-3085/$ – see front matter © 2008 Published by Elsevier B.V. on behalf of The Institution of Chemical Engineers. doi:10.1016/j.fbp.2007.10.019

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

can be hazardous to the process, because they may introduce contamination when used, or placed in an improper manner giving unrepresentative information on the production hygiene. The desired result of hygiene testing is to detect if the amount of remaining spoilage and pathogenic microbes on a cleaned process surfaces is at acceptable low level. The presence of microbes on surfaces after cleaning procedure can be detected from samples taken by swabbing, swiping or pressing a device onto the surface. The reliability of results from such microbial analysis is primarily dependent on how representative the sample is compared to the situation at the surface tested and on the accuracy of the method applied. Traditional methods are typically based on culturing of the microbes removed from the surface providing information about the amount of microbes and if selective growth media is used also presumptive knowledge about the microbe strain is obtained. If more information is needed from microbes contaminating surface samples can be analysed using methods based on molecular biology. Modern molecular biology provides several accurate methods to analyse microbes from samples (Maukonen et al., 2003). Still the sample needs to be representative in order to benefit the information obtained using molecular biology methods. For instance tracing a pathogen contaminant by comparing fingerprints of the microbes on various samples does not necessarily lead to the source of contamination since sampling of surfaces is improper. Most DNA-based identification methods destroy the living cells in order to get the DNA from cell to analysis. Sampling for these types of analysis does therefore not need to be gentle for cells but on the other hand cells that have been dead before sampling took place will be analysed along with the living cells and once the microbe is identified it cannot be cultured for further analysis. Cells inactivated before sampling due to disinfection or heating will give similar signals as living cells unless separated from the viable cells. The sampling methods have limitations and epifluorescent microscopy studies have clearly showed that only a small part of biofilm can be removed with swab from surface (Salo and Wirtanen, 1999). The microbe yield with swabs and contact agars from stainless steel surface soiled with known amounts of microbial soil left to dry for 5 min was max 20% according to a collaborative comparison test (Salo et al., 2000, 2002). The amount of viable but non-culturable microbes can also be significantly high (Maukonen et al., 2000). Martiny et al. (2001) has reported that less than 1% of microbial population is culturable under oligotrophic conditions and for instance denaturing gradient gel electrophoresis (DGGE) could provide a good profile of the community present in water samples. The cleanability of especially large tanks can be validated utilising visual inspection. The visualization can be made more effective by soiling the tank with soil containing a fluorescent component, for instance uranine or riboflavin, before performing the cleaning procedure. The areas with fluorescent soil debris becomes visible by using UV-light (Kold and Skræ, 2004). Though it occasionally can be difficult to determine, if areas light up due to reflections in the surface material, imperfections in the surface finish or actual fouling as intended. The impurities are easily identified on stainless steel surfaces whereas plastic material itself often is fluorescent. Thus an experienced observer is needed in this case to separate fluorescence soil from fluorescence of plastic sur¨ as, ¨ 2006). Generally the fouling appears as a dim face (Forn white and occasionally pink coating. Jacob and Brandl (2002)

205

also noticed limitations in the visual inspection when they monitored the efficiency of the cleaning and disinfection procedures in tanks in breweries. They have used UV illumination to detect deposits. They also have used Fuchsine staining to detect proteins and Lugol’s iodine solution to detect starch. Such methods are commercially available from, e.g. Bactoforce. There are at present no acceptable methods for directly correlating numbers of bacteria in the water running in system with the number in biofilms on surfaces of the system (Camper and McFeters, 2000). Despite this water samples have been collected and analysed for predicting the presence of biofilms in drinking water distribution systems because of the limitations in access to surfaces for sampling biofilm (Camper and McFeters, 2000). Side streams derived from the process line and pilot devices imitating real distribution system have been used in both food processing and drinking water distribu˚ tion systems to obtain biofilm samples (Storgards et al., 1999; Camper and McFeters, 2000). These samples provide rough estimation of biofouling in the real entire distribution system. Studies show that pilot plant tests can be used to imitate the cleaning efficacy of chosen parameters towards artificially soiled surfaces representing possible worst-case situations (Wirtanen, 1995). Hence such down scaling of equipment will be applied in the present study. As mentioned earlier the cleanability results can be used for optimising the cleaning parameters and plant design. Hence ways to relate information on cleaning efficiency to the physical design of processes must be available. Process designers apply various computational tools optimise shapes in relation to stability of constructions and flow of products through plants. This has lead to simulations obtained using computational fluid dynamics (CFD) which can be linked to cleaning efficiency from pilot plant experiments (Friis and Jensen, 2002a,b). CFD is used in many applications to model the bulk parameters of fluid flows. Recently model developments have made it possible to resolve local flow phenomena in specific positions and near walls, which is of interest in the study of cleaning processes (Jensen, 2003). Results of the CFD simulations yield information about wall shear stresses and the flow rates in different parts of the system. A combination of wall shear stress, fluid exchange and turbulence conditions can be used to predict areas that are not properly cleaned in both simple and complex flow systems (Friis and Jensen, 2002a). Jensen (2003) has made CFD simulations for the prediction of the hygienic design of valves, pipes, etc. In the CFD modelling tool the flow phenomena, e.g. wall shear stress and fluid exchange, are used for predicting the cleanability in the flow systems. The combination of surface topography, fluid dynamics and surface microbiology provides a good basis for defining the proper hygiene sampling locations in closed processing systems. However at present the effect of surface topography cannot be included in the CFD simulations. The methods used for detecting problems in tank cleaning must be practical, informative, and harmless to the tank surface as well as to the product since residues of sampling chemicals can always remain on surfaces. In this study cleaning of soiled tank has been performed using only cold water and by observing the cleaning result during cleaning procedure. The aims of this study were to find appropriate methods to detect cleanliness in pilot scale tanks, to find out which parts of the tank were difficult to clean and to optimise the cleaning procedure. This pilot scale study was performed to

206

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

improve the investigation of cleanability of process scale tanks and to choose suitable methods and sampling areas for use in process scale evaluations. Furthermore interpretation of data from various sampling methods is needed in cleanability assessment.

2.

Materials and methods

2.1.

Equipment

The test equipment was a stainless steel tank with a volume of 80 l. The inner diameter was 400 mm and the height 800 mm (Fig. 1). In the tank lid there was connections, e.g. for a spray ball. The total length of spray ball was 145 mm and the rotating tip was 50 mm in diameter. The spray ball was connected to the lid with 140 mm long pipe.

2.2.

Soil

Cleaning tests were performed in the tank soiled with sour milk (0.5 l økologisk tykmælk, Thise Dairy, Denmark) containing Bacillus stearothermophilus spores (VTT E-88318). The clean tank was painted with the milk-based soil using a painting brush. Since it is important to make the soiling as repeatable as possible the whole 0.5 l of sour milk mixed with spores was used for each test and spreading was performed as evenly as possible. The soiled tank was left to dry for 3 h. It was placed under a hood. The fingers were not moistened by the dried soil when it was touched lightly. Another type of soil was obtained when the tank was filled with beer and kept at room temperature for 2 weeks. The tank surfaces were thus soiled with a natural beer-based soil containing yeast cells.

2.3.

Cleaning procedure

The tank was connected to a system which carried out two different types of cleaning procedures. The first mild cleaning procedure in which water was pumped from a water vessel to the tank and the flow rate was measured with a flow meter before the water went to the rotating spray ball connected through the lid of the tank. The duration of the mild cleaning procedure was 20 min. A more severe cleaning procedure was adapted from the guideline on testing cleanability of closed equipment published by European Hygienic Engi-

neering and Design Group (EHEDG). This cleaning procedure consisted of 1 min pre-rinsing with cold water, 20 min alkali cleaning performed by circulating 60 ◦ C 1% EHEDG test cleaner PD332 (Lever Industrial, The Netherlands) liquid in the system and rinsing with cold water for 1 min (Timperley et al., 2000). The velocity of cleaning liquid in the pipeline before spray ball was adjusted to 1.5 m/s in both cleaning set-ups. The flow rate of the fluid entering the tank through spray ball was 2200 l/h. The thickness of the falling film was uneven since the amount of liquid distributed on the perimeter of the tank is unlikely to be constant. The water was removed from the test tank using an additional pump in the bottom of the tank.

2.4.

Visual analysis

In order to improve visual observation of the effectiveness of the cleaning procedure fluorescent chemicals were applied to the surface before cleaning. In test set-ups four different spreading techniques were used and they were spraying the tank wall with 0.012% uranin (fluorescent sodium salt, Merck, Germany), spraying the tank wall with 0.1% riboflavin (B2 vitamin, BioChemika, Germany), painting the tank wall with sour milk containing 0.012% uranin or adding 1.0 l water containing 1.0 g riboflavin in the tank containing beer-based soil. The fluorescence was visualized using a UV-lamp (Lamag, Merck) with wave length 366 nm and Dark Reader lamp (Clare Chemical Research, USA) with orange detection glasses.

2.5.

Sampling

Samples for determination microbial load on the tank surface were taken from five different sampling points, which were the upper part of the tank wall, the middle part of the tank wall, the lower part of the tank wall, the stainless steel part of the lid and the glass window part of the lid. In case of mild cleaning samples were taken before cleaning and after 1, 2, 5, 10 and 20 min cleaning (cleaning was interrupted for sampling and it continued immediately after the sampling) and in the strong cleaning procedure samples were taken before and after cleaning. Microbial sampling was performed with swabs followed by conventional culturing and ATP measurement, with commercial contact agars and with self-made contact agars.

Fig. 1 – The test equipment was a stainless steel tank with a volume of 80 l, with an inner diameter of 400 mm and a height of 800 mm. (a) Tank opened for sampling, (b) sampling the lid with contact agar, and (c) sampling the tank wall with contact agar.

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

2.6.

Culturing

In the detection of microbes contact agar methods and conventional culturing was utilised. The methods were commercial contact agar Petrifilm AC (3M, USA; moistened with 1 ml physiological saline for 2 h before use) and Hygicult® TPC (Orion Diagnostica Oy, Finland) and self-made Nutrient Agar (NA; Merck, Germany) contact agars. They were used by pressing agar towards tested surface for approximately 3 s. Swabbing was performed using cotton tipped swabs moistened with physiological saline for collecting microbes from approximately 100 cm2 . Microbes were transferred into 5 ml physiological saline by vortexing the test tube containing the swab properly. Alternatively sterile non-woven cloths were used for sampling by first moistening them with sterile physiological saline, swiping the sample area wearing sterile gloves and mixing the non-woven cloth with 10 ml physiological saline in sterile stomacher bag using a homogenizator for 30 s. Samples were cultivated on NA using pour plating and spreading techniques. The contact agars and Petri dishes with B. stearothermophilus spores were incubated at 58 ◦ C for 1 day and with yeast cells at 25 ◦ C for 4 days.

2.7.

Microscopying and agar moulding

Six stainless steel plates sized 2.5 cm × 7.5 cm were attached to the tank wall using double-sided tape before the soiling process in order to get suitable samples for microscopying and agar moulding. Detached surfaces were stained with acridine orange (Becton & Dickinson, Franklin Lakes, USA) for 2 min and rinsed with tap water. After that the stainless steel plates were analysed using epifluorescence microscopy (Nikon Eclipse E1000, Nikon Instech Co. Ltd., Kawasaki, Japan) combined with image analysis system (Image Pro plus 5.1, Media Cybernetics Inc., Silver Spring, USA). WL Nutrient agar (WLNA, Oxoid, UK) was poured on the detached plates and after agar solidification the samples were incubated at 25 ◦ C for 3 days.

2.8.

Organic debris

Samples for ATP detection were taken with ATP-free sterile swabs (ThermoLabsystems) moistened with physiological saline from the top, middle and low parts of the tank wall and from the stainless steel surface and the glass window in the lid. The ATP test was performed by mixing the swab with 0.3 ml Hygiene ATP Releaser (ThermoLabsystems) in a cuvette for 20 s followed by an addition of 0.3 ml Hygiene Monitoring Reagent (ThermoLabsystems) and thereafter the light output was measured with Kikkoman Scil Lumitester PD-10 (Chiba, Japan). Also a ready-to-use ATP reagent kit Lucipac W (Kikkoman) was used. Furthermore, the test kits Cleantrace ATP-pen (Biotrace, UK) and ATP hygiene kit HS (Biothema, Sweden) were used. The measurements were performed according to instructions given by the manufacturer.

2.9.

Chemical residues

Information about chemical residues in rinsing water was obtained by collecting samples from the final rinsing water from a sampling tap every 10 s during 1 min rinsing period. The toxicity of the water samples were analysed using BioTox kit (Aboatox, Turku, Finland). Test was performed by measuring the light output of 0.5 ml Vibrio fisheri bacteria suspension with

207

PD-10 luminometer, adding 0.5 ml sample, mixing the cuvette content and incubating it at room temperature for 5 min. The light output was measured and the inhibition was counted comparing light output for normal tap water and rinse water sample.

3.

Results and discussion

The study provides useful information about behaviour of fluorescent chemicals. The beer-based soiling liquid containing riboflavin was extremely fluorescent during the whole 2 weeks soiling period, but the tank surfaces did not fluoresce after emptying even though the surface was visually dirty. Visual observation of cleanability using fluorescent stain was best when used together with the sour milk soiling, since the stained sour milk attached more firmly to the surface compared to when fluorescent dye sprayed on the surface alone. The cleaning procedure for the tank soiled with sour milk was too effective to be able to detect differences clearly by eye; only pipeline connections in the lid had visible residues. It is evident that this test procedure needs to be optimised for each application, since the soil affects the spreading of the indicator stain. Furthermore, the low power at the Dark Reader lamp forced the observations of the surface to be performed from a close distance, which made the observation of the tank bottom impractical. The advantages in using a Dark Reader lamp compared to UV-lamp are that it is harmless to the user, it is also small in size and it is working on battery. However, the UV-lamp was much more effective in visualizing fluorescence; however it was too big to be moved easily inside the tank. The determination of cleanliness after the CIP procedure is a difficult task especially since cumbersome spots needs to be observed. The most practical way of finding these hot spots are to soil the surface with a fluorescent indicator before performing the cleaning procedure and observe residues visually after cleaning. Microbial detection based on swabbing and culturing is laborious and furthermore the area of the swabbed surfaces was difficult to measure accurately in this study, because the sampling area in the lower parts of the tank was not easy to reach and see properly. Due to the difficult and unstable working position the swabbing pattern and force used for pressing the swab against the surface varied. The use of sterile form aiding in sampling of known area was impossible to use especially near the tank bottom and on the lid with its connections. The above-mentioned uncertainties made the culturing results unclear and no logic trends of microbes left on the surfaces after cleaning could be drawn. Sampling using sterile gloves and non-woven cloth made it possible to use more force in sampling since the wooden stick on the cotton tipped swabs broke when too high force was used in swabbing. The sampling was not performed by wetting the surfaces with a sampling aid called Spraycult (Orion Diagnostica) since according to earlier studies no major advantages can be obtained on already wet surfaces (Tuompo et al., 1999). The most practical sampling method for detection of microbes on the tank surfaces was the contact agar method. Two commercial contact agars were used. Petrifilm was more flexible than Hygicult® . The agar in the Hygicult® was moulded on a rigid plastic frame and thus it did not always have proper contact with the often curved surface. The self made contact agar had same disadvantages as Hygicult® . Fig. 2

208

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

Fig. 2 – Comparison of results obtained using Petrifilm and Hygicult® to detect cleanliness of milk and beer soiled tanks after cleaning with EHEDG test procedure. Before cleaning the soiled tank surfaces contained more than 25 cfu (colony forming units)/cm2 on every sampling place according to the Petrifilm method. According to the Hygicult® method the microbial load on these sampling places varied from less than one to more than 100 cfu/cm2 . The Petrifilms containing over 500 cfu/20 cm2 were interpreted as too numerous to count and marked as 500. The growth on Hygicult® was estimated using a model chart provided by the manufacturer.

shows the results from the tank surfaces cleaned with the EHEDG test cleaning procedure using both commercial contact agars. The standardization of force used in taking contact agar samples can improve the repeatability of the results. For self made contact agars a Count-tact applicator (bioMerieux Inc., USA), which is a sampling tool pressing the agar towards surface with known pressure and which gives a voice signal after 10 s contact time, can be used on straight surfaces. This tool needs to be modified for sampling with Petrifilms or other similar flexible contact agar applications on curved surfaces. The contact agar method can be recommended to be used in detecting the cleanliness of tank surface after cleaning because the results correlate well with swabbing results when both methods are used in repeatable test set-up (Salo et al., 2000, 2002). In Fig. 3 the results obtained with Petrifilms from the mild cleaning procedure on the tank soiled with sour milk are shown. The lower parts of tank wall were the most difficult to clean using the mild rinsing procedure. The upper parts of tank wall as well as the smooth parts of the lid were the best cleanable parts. The same trend in the results was obtained with cleaning test on the tank soiled with beer (results not shown). The bottom of the tank soiled with beer did not show

Fig. 3 – Results from follow-up of a 20 min rinsing of a tank with sour milk-based soil using Petrifilm contact method.

any signs of progressed cleaned during 20 min rinsing period, where as the top wall became a bit cleaner after 1 min of cleaning and the middle part of the wall was cleaner after 10 min cleaning. The more severe cleaning procedure containing cleaning agent cleaned the bottom areas better since the detergent had had time to soak soil due to insufficient removing of the cleaning liquids from the tank (Fig. 2). The pump removing the cleaning liquid from the tank should be improved for instance by adding an eductor (Tri-Clover Inc., Kenosha, USA) to the system to be able to empty the tank effectively without avoiding that the pump runs without liquid. The approach of attaching coupons of stainless steel to the tank wall needs to be improved. The majority of the small steel plates intended for microscopying and moulding purposes did detach during the soiling and cleaning procedure. This method can only be used in pilot plant studies, since loose stainless steel plates in an industrial process plant may harm the process or the product. Additionally, the gluing of these coupons to the tank wall inevitably leads to an unhygienic construction. It may be questioned if the steel plates experience flow conditions similar to the remaining part of the wall. An attached plate of 1 mm thickness will cause an unknown disturbance in the flow of the falling water film. However, the results obtained from stained stainless steel plates using microscope and image analysis are very informative revealing area fraction covered with microbes and other debris. No conclusions concerning the cleanliness of various areas of a tank can be made based on microscopying, because of the detached plates. The lack of a proper test surface series was also the problem with the moulding method. However, WLNA poured on and microscopying of individual detached plates showed microbial residues on the plates. These results indicate qualitatively that the applied cleaning procedures did not remove all foreign bodies on the tank wall. Unfortunately no quantitative result has been obtained in these experiments but this will be possible following optimisation of the method. The humidity in the surrounding of stainless steel plates during incubation needs special attention since thin layer of agar easily dries and becomes useless for growing microbes. Visual inspection of surface reveals valuable information about debris left on surface. Though these methods are difficult to use as routine control methods during production it is recommendable to use these methods for interpretation of results obtained with more practical methods like swabbing and contact agar methods. The ATP method was used for measuring organic load including microbes from swabbed samples. However, the standard deviation of results from parallel samples was not acceptable. These results did not reveal differences between cleaned areas nor cleaning procedures. There are various explanations for unsuitability of ATP method for this study. The sensitivity of ATP method is often too low for detecting small amounts of microbes from cleaned surfaces. It has also been noticed that some residues of cleaning agents and disinfectants disturb the enzyme reaction by either inactivating or over activating the light production (Lappalainen et al., 2000). Furthermore, the ATP level in spores is very low. Sampling using swab for ATP measurements had the same weaknesses as the culturing method, i.e. variations in surface area and swabbing force. Methods based on microbial growth do not give results immediately whereas the main advantage of ATP method is that it provides rapidly results. The speed of the ATP method was not valuated too high in the pilot plant studies. The ATP method is at its best in routine control, when

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

Fig. 4 – Chemical residues in rinsing water samples.

same sampling places with standardized sampling area can be sampled regularly and variations of results can be connected to possible actions. According to Paez et al. (2003) a commercial ATP bioluminescence system can be used to evaluate the cleanliness of milking equipment, bulk tanks and milk transport tankers. Odebrecht et al. (2000) studied the applicability of bioluminescence methods for testing hygiene in beer fermentation and maturation tanks and yeast tanks in breweries. Results indicated that the bioluminescence method was not suitable for the detection of microbial contamination, as the results did not correlate with those obtained by conventional microbial culturing techniques. The presence of chemical residues originated from the final rinsing water of the EHEDG test cleaning procedure. The results obtained with the photobacteria-based method showed clearly that chemical residue exists in the rinsing water throughout the 1 min rinsing period. After 30 s rinsing a reduction of the chemical residues could be detected with this sensitive method (Fig. 4). In order to get more precise qualitative information about the chemical residues, samples needs to be diluted before measurements and standard curves of used cleaning agent should be performed. This method has been developed to be easily used only for controlling when final rinsing water is free from chemical residues. It would be desirable to predict the proper location of sampling spots. Hence, we turn to CFD which has proven to be applicable to relate flow characteristics to cleaning efficiency in closed process equipment. Combining knowledge on fluid dynamics and microbiology and soil provides an excellent basis for improvement of the hygienic design. At least qualitative experimental results from cleanability tests performed with pilot scale equipment are important since a sound basic knowledge is required in order to interpret the CFD simulation results properly. This can be used for predicting cleanability of this type of process system. The validation of the cleaning procedure and the design of a proper cleaning system can be supported and improved using CFD but the reliability and suitability of CFD simulations for estimating and improving tank cleaning still needs further studies. Such information is not available today but we believe that the results from our work will provide a good starting point for combining CFD simulation results and traditional cleanability sampling. We have approached the interpretation of cleanability sampling by means of CFD using a simplified test set-up simulating inclined bottom of tank (Salo et al., 2006). Accord-

209

ing to these studies the designing of efficient and economical cleaning procedure should also be based on an analysis of flow actions including optimisation of thickness of the liquid layer. It is recommended to use more than one method for analysing the cleanability of tanks since all the detection methods have limitations. By combining the knowledge obtained using various methods the problems in cleanability of tanks can be revealed and understood more deeply. The areas with cleanability problems in large tanks can be localised using methods based on visual observation and studying the hygienic design. This information can be utilised when choosing the sampling areas for microbial analysis. As an example of relying in one method in interpreting cleanability in our test set-up with the pilot scale tank the culturing-based methods can give misleading information due to inadequate rinsing. In such cases at least two methods are needed to avoid incorrect information. The culturing of microbes from the bottom area and testing of chemical residues from rinsing water as well as observing the debris with microscopy support interpretation since chemical residues may inhibit the microbial growth and basic microscopying does not separate dead and viable cells.

4.

Conclusion

Good hygiene is important in equipment used for food processing. The fermentation and storage tanks in dairies and breweries can be up to several hundred cubic meters with just one spray ball mounted in upper part of the tank. According to our studies the bottom of these types of tanks is the most difficult area of the entire tank to clean. The enormous size of the tanks and lack of proper sampling spots makes sampling of tanks using conventional swabbing method problematic. Therefore improved sampling procedures are needed in evaluating the cleanliness of tanks after CIP procedure. Since it is laborious to take samples and observe the cleanliness of whole surface area of large-scale tanks it is recommended to extend the application of predictive computer simulations as a tool in cleanability assessments. This study showed that microscopying, moulding and ATP methods were not practical methods for studying cleanability of tank. Visual observation of the cleaning efficiency aided with UV-light in large-scale equipment is an important practical method which requires substantial user experience to be valuable in the cleanability assessment. However visual observation with UV-lamp is not relevant method in laboratory research due to lack of quantitative information. The traditional culturing methods are used especially when information about microbes causing contamination is needed. When interpreting the results it must be taken into account that sampling using swabs is limited to reachable areas and also that a poor recovery decreases the representativeness of the sample. The contact agar method is similarly based on microbial growth and according to our study it revealed more logical results than culturing of swabbed samples.

Acknowledgements Food technologists Helle Mathiasen and Preben Bøje Hansen are acknowledged for assisting the trials in pilot hall at BioCentrum-DTU. These studies were partly connected with the Nordic cooperation project “DairyNET—Hygiene Control in Dairies” (Wirtanen and Salo, 2004). The Nordic Innova-

210

f o o d a n d b i o p r o d u c t s p r o c e s s i n g 8 6 ( 2 0 0 8 ) 204–210

tion Centre has been funding this project which is gratefully acknowledged.

references

¨ Anon., 2003, Hygienialain mukaisten naytteiden ottaminen, ¨ ¨ Asetus nro 9/EEO/2003, Maa-ja metsatalousministeri o, www.mmm.fi/el/laki/i/i10, November 26, 2004, p. 10 (in Finnish). Camper, A. and McFeters, G., 2000, Problems of biofouling in drinking water systems, in Industrial Biofouling. Detection, Prevention and Control, Walker, J., Surman, S., & Jass, J. (eds) (John Wiley & Sons, Chichester, UK), pp. 15–38. Friis, A. and Jensen, B.B.B., 2002, Hygiene and flow modelling in food processing equipment, in Fouling, Cleaning and Disinfection in Food Processing, Wilson, D., Fryer, P., & Hasting, A. (eds) (University of Cambridge, Cambridge, UK), pp. 119–126. Friis, A. and Jensen, B.B.B., 2002, Prediction of hygiene and flow modelling in food processing equipment. Trans IChemE, Part C, 80: 281–285. ¨ as, ¨ R., 2006, UV-lamppu helpottaa elintarviketeollisuuden Forn hygieniavalvontaa. Kehittyva¨ Elintarvike, 17(3): 42–43, in Finnish Jacob, F. and Brandl, A., 2002, Auffinden von Schwachstellen in Reinigungssystemen mit UV-Licht. Brauwelt, 142: 1162–1163. Jensen, B.B.B., (2003). Hygienic Design of Closed Processing Equipment by Use of Computational Fluid Dynamics. (Technical University of Denmark, Lyngby, BioCentrum-DTU). Kold, J. and Skræ, A.M., 2004, A method for assessing the cleanability of open processing equipment, in DairyNET—Hygiene Control in Nordic Dairies, Wirtanen, G. and Salo, S. (eds) (Otamedia, Espoo, Finland), pp. 213–216. ¨ Lappalainen, J., Loikkanen, S., Havana, M., Karp, M., Sjoberg, A.-M. and Wirtanen, G., 2000, Microbial testing methods for detection of residual cleaning agents and disinfectants-prevention of ATP bioluminescence measurement errors in the food industry. J Food Prot, 63: 210–215. Martiny, A., Albrechtsen, H.-J., Arvin, E. and Molin, S., 2001, Application and strategy for identifying and monitoring bacterial fouling, in Proceedings of 32nd R3 Nordic ˚ Symposium, Mansson, L. (ed) (R3 Nordic, Genarp, Sweden), pp. 354–364. Maukonen, J., Mattila-Sandholm, T. and Wirtanen, G., 2000, Metabolic indicators for assessing bacterial viability in hygiene sampling using cells in suspensions and swabbed biofilm. Lebensm Wiss Technol, 33: 225–234. ¨ o, ¨ J., Wirtanen, G., Raaska, L., Maukonen, J., Matt Mattila-Sandholm, T. and Saarela, M., 2003, Methodologies for

the characterization of microbes in industrial environments: a review. J Ind Microbiol Biotechnol, 30: 327–356. Odebrecht, E., Schmidt, H.J. and Franco, B.G.M., 2000, Studies on applicability of bioluminescence in the brewery. Comparative studies and critical evaluation. Brauwelt, 140: 1904–1905, 1908–1915 Paez, R., Taverna, M., Charlon, V., Cuatrin, A., Etcheverry, F. and da Costa, L.H., 2003, Application of ATP-bioluminescence technique for assessing cleanliness of milking equipment, bulk tank and milk transport tankers. Food Prot Trends, 23: 308–314. ¨ Salo, S., Alanko, T., Sjoberg, A.-M. and Wirtanen, G., 2002, Validation of the Hygicult® E dipslides method in surface hygiene control: a Nordic collaborative study. J AOAC Int, 85: 388–394. ˚ Salo, S., Johnsen, S., Jensen, B.B.B., Storgards, E. and Friis, A., 2006, The mechanical effect of rinsing in cleaning of fermentation tank bottoms, in Fouling, Cleaning and Disinfection in Food Processing, Wilson, D.I., Chew, J.Y.M., Fryer, P.J., & Hasting, A.P.M. (eds) (Department of Chemical Engineering, University of Cambridge, Cambridge, UK), pp. 162–171. ¨ Salo, S., Laine, A., Alanko, T., Sjoberg, A.-M. and Wirtanen, G., 2000, Validation of the microbiological methods Hygicult dipslide, contact plate, and swabbing in surface hygiene control: a Nordic collaborative study. J AOAC Int, 83: 1357–1365. Salo, S. and Wirtanen, G., 1999, Detergent based blends for swabbing and dipslides in improved surface sampling, in Biofilms: The Good, the Bad and the Ugly, Gilbert, P., Branding, M., Walker, J., & Wimpenny, J. (eds) (Bioline, Cardiff, UK), pp. 121–128. ¨ ˚ Storgards, E., Simola, H., Sjoberg, A.-M. and Wirtanen, G., 1999, Hygiene of gasket materials used in food processing equipment. Part 1. New materials. Trans IChemE, Part C, 77: 137–145. Timperley, A.W., Bourion, F., Benezech, T., Carpentier, B., Curiel, ¨ G.J., Haugen, K., Hofmann, J., Kastelein, J., Ronner, U., Tragardh, C. and Wirtanen, G., 2000, A method for the assessment of in-place cleanability of food processing equipment, European Hygienic Equipment design Group (EHEDG) Document No. 2, p. 13. Tuompo, H., Salo, S., Scheinin, L., Mattila-Sandholm, T. and Wirtanen, G., 1999, Chelating agents and detergents in sampling biofilm, in Biofilms: The Good, the Bad and the Ugly, Gilbert, P., Brading, M., Walker, J., & Wimpenny, J. (eds) (BioLine, Cardiff, UK), pp. 113–120. Wirtanen, G. 1995, Biofilm Formation and its Elimination from Food Processing Equipment (VTT Publications 251). Espoo, Finland: VTT Offsetpaino 106 p. Wirtanen, G. Salo, S., 2004, DairyNET—Hygiene Control in Nordic Dairies (VTT Publications 545). Espoo, Finland: Otamedia. p. 253.