domnick hunter - a division of Parker-Hannifin Corporation. World-Class Filtration, Purification and Separation Solutions to Industry

• View all products
• Customer Support
• Compressed Air Technical Centre
• Global Exhibitions and Events
• Careers
• Awards & Accreditations
• Legal

by David Porter, CEng
Engineering Manager, domnick hunter ltd. Abstract :-

This paper discusses the issues of protecting the quality of carbon dioxide used for serving draught beers at the point of sale. There have been well publicised incidents of contaminated carbon dioxide (CO₂) entering the supply chain and affecting the quality of carbonated beverages. These contaminants may affect taste, odour, the appearance of foam head and presentation of the beer or may be controlled by regulation to prevent health issues. An in-line purifier is presented which has been designed to act as a safeguard against possible CO₂ contaminants and maintain the qualities of the beer as the brewer intended.

Key Words :-
CO₂, contamination, purification, draught beverage dispense.

Background :-
In response to a growing awareness in the soft drinks industry that carbon dioxide should be treated as an ‘ingredient’ of bottled or canned products, domnick hunter developed a CO₂purifier for use in the packaging plants. The aim is to provide in-line protection acting as trace contaminant removal devices [1].

Figure 1: Bottling Plant Scale CO₂ Purifier These products, in conjunction with a rigorous quality control (QC) program, ensure that no contaminated gas comes into contact with the beverage. If an out of specification batch of CO₂ is delivered to the plant, the purifier protects beverage production whilst the quality system and gas monitoring program detect and remedy the CO₂ quality excursion.

Figure 2 :- CO₂ chain of supply to bottling/canning plants.

It can be seen that there is a short chain of supply between the CO₂ producer and the bottling / canning plant which helps ensure the cleanliness and quality control of the CO₂. Due to the large volumes of gas used and centralised geographical locations, the investment in sensitive measuring equipment and skilled operators to monitor gas quality can be justified. This results in a very robust QC regime. However, a completely different situation applies to CO₂ gas used for draught beer dispense which is characterised by :-

• A very large number of physical locations,
• No on-site gas monitoring / quality inspection equipment,
• Bar staff focused on customer service rather than equipment maintenance / QC activities
• Lower gas volumes, flow rates and operating pressures.

This supply chain is illustrated in Figure 3.

Figure 3 :- CO₂ chain of supply for draught beer applications.

The rationale for a point-of-sale trace gas contaminant removal device is the same as for those fitted at the brewery / bottling plant, namely that the purchased CO₂supply may be of varying quality. If any impurities are present they may affect the appearance and taste of the beer. In fact, certain contaminants may present a risk to the consumers health. To combat these problems, a small scale protection device, fitted into the draught beer CO₂ supply line, has been developed.

CO₂ Contaminants :-
Over the last 2 years, bodies such as the International Society of Beverage Technologists (ISBT) and European Industrial Gas Association (EIGA) have written guidebooks [2,3] which include specifications for maximum contaminant levels for beverage quality CO₂ and approved analytical test methods. Typical named contaminants include :- Volatile Organic Hydrocarbons
Aromatic Hydrocarbons (eg benzene, toluene, cyclohexane)
Sulphur containing compounds (eg hydrogen sulphide, carbonyl sulphide)
Acetaldehyde
Non-Volatile Organic Hydrocarbons (eg grease)

The maximum concentration of each group of contaminants may be motivated be either their effect on taste or smell of the beverage, beverage quality control (eg effects on other ingredients or degradation of shelf life) or by regulatory measures to protect consumer’s health. For example, the maximum volume concentration of aromatic hydrocarbons is set at 20 parts per billion. The primary source of the contamination is the chemical process used to manufacture the CO₂. Carbon dioxide is an unusual industrial gas in that it is commonly produced commercially using a wide variety of production methods. Each production process uses differing source chemicals and tends to leave different trace residuals which then become considered as contaminants [3]. A secondary source of contaminants must also be considered, namely the gas bottles and distribution system. As indicated in Figure 3, the gas bottles move between the points of use (licensed premises) and the regional bottle filling companies. There are very large numbers of CO₂ bottles in circulation and guidelines for bottle handling and filling exist [4] but are not always complied with. An typical example of bottle mishandling is that of an empty bottle returned to the filling plant with the valve left open allowing atmospheric contamination and moisture to enter the bottle. (The guidelines require a minimum positive pressure to be maintained to prevent this ingress.) Water vapour collecting and condensing in the cylinder can combine with the CO₂ to form carbonic acid which may attack and corrode the cylinder. As CO₂is frequently supplied to licensed premises in gas bottles, a ‘quality incident’ is defined as a single full gas bottle containing out of specification CO₂. A typical gas bottle used for CO₂ distribution contains approximately 50 lbs. of gas. The test work described below will reference test contamination levels to the ISBT CO₂ specification and will quote the equivalent number of gas bottles successfully processed with reference to a typical 50 lbs. gas bottle. Product Concept :- With a large number of gas cylinders moving over large geographical areas from a multiplicity of filling locations to the end users, effective bottle control would be very difficult to implement logistically. An alternative approach taken here is to install a protective device at each licensed premises to remove contaminants should they be present and provide ‘point of use’ protection (see Figure 4). This device would be sized for the maximum CO₂ flowrate required by the retailer to serve draught beer and have an in-service life so that frequent maintenance would be avoided. Due to the wide range of contaminants potentially present in the CO₂, three specially selected adsorbants were used, each having different adsorption characteristics. The resulting composite bed generates a very robust protection against the wide range of expected contaminants.

Figure 4:- Product Concept

Verification Testing :-
To verify the performance of the CO₂ purifier unit a series of tests were undertaken :- 1) Verification of contaminant removal capability
2) Endurance testing; challenge the adsorbant bed for the equivalent of several gas bottles of contaminated CO₂
at rated conditions and monitor inlet and outlet concentrations.
3) Contaminant retention; pass clean CO₂ through the adsorption unit and verify that the contaminants do not desorb and contaminate an otherwise clean CO₂ supply. The verification of the ability of the adsorbants to remove the contaminants of interest was conducted using small quantities of adsorbant and specially produced challenge gas mixtures. Several adsorbant materials were screened for the application with the most suitable eventually selected. This work was carried out as part of the early product design and testing undertaken by domnick hunter and is not reported here . Specially produced mixtures of CO₂ and trace contaminants were obtained from a specialist gas company. These were then mixed in various dilution ratios with clean CO₂ to obtain the target inlet concentrations of the contaminants. The dilution ratio of clean and contaminated CO₂ was controlled by 2 mass flow controllers; by adjusting the relative flow rates of clean and contaminated gases, a variety of dilution ratios, and hence inlet concentrations, could be produced. The mixed gas was then passed through the adsorption device with inlet and exhaust contaminant levels monitored for impurities. The experimental set up is illustrated in Figure 5. The test conditions were as follows :- Flow :- maximum rated flow, 1.2 lbs/hr (5 l/min ANR)
Pressure :- nominal working pressure, 15 psig (1.1barg)
Temperature :- room ambient, 70°F (21°C)

Figure 5 :- Experimental Test Rig

The contaminated CO₂was allowed to pass through the purifier for a period of time and gas samples were taken periodically, typically every 24 hours. The equivalent number of 50 lb gas bottles treated during the tests are given in the result summary table. This enabled data regarding continued contaminant removal over time to be gathered. After this period, the contaminated gas was removed and only clean CO₂ allowed to pass through the purifier. Inlet and outlet gas samples were again taken to confirm that the contaminants removed in the first part of the test were not released to contaminate the otherwise clean CO₂ flow. This was allowed to run for a further period and would demonstrate that previously adsorbed contaminants were not desorbed. The gas samples were analysed using one of two methods depending on the contaminates of interest. Volatile organic carbons were sampled by adsorption tube followed by thermal desorption into a gas chromatograph fitted with mass spectrometer and flame ionisation detectors. The acetaldehyde was analysed by bubbling the CO₂ through solutions of iso-octane / DNPH. Pairs of bubblers were used with the amount of contaminant quoted in the results being the sum of contaminant found in the front and rear bubbler solutions. Exposure times and gas flowrates were selected to give a sufficiently low detection level to give reliable results at expected contamination levels. These were typically 100 cc/min sample flows for 100 minutes giving a total sample volume of 10 litres of CO₂. All contaminant sample analysis was carried out by an external, accredited, analytical chemical laboratory. Results :-

The acetaldehyde results are favourable. The inlet concentrations were approximately 15 times those specified by ISBT and although outlet levels are seen to rise towards the end of the test, they remain approximately 1/8 of the ISBT requirement. The acetaldehyde contamination was removed after processing the equivalent of 80% of a 50 lb bottle. The ethanol results are very positive with inlet levels of 25 ppmV reduced to sub 0.002 ppmV levels, a reduction of over 1000 times. The inlet concentration is compared with the total allowable contamination for volatile hydrocarbons (excluding methane) set by ISBT at 20 ppmV. The switch from contaminated to clean CO₂ streams did not cause any detectable release of previously adsorbed ethanol. In fact, all but one result from the unit outlet is quoted as the detection level of the analysis method as no measurable amount of ethanol was observed. The equivalent of 2.5 bottles of contaminated gas was treated during the test, with approximately 1.5 bottles of clean gas processed with no indication of contaminant desorption.

The benzene results are favourable. Inlet concentrations of approximately 0.4 ppmV are approximately 20 those specified by ISBT. Outlet levels were very low and approximately ¼ of the ISBT specification and only one measurement was greater than the detection level of the measurement process. A reduction in Benzene concentration of approximately 80 times was observed. The benzene contaminant was tested simultaneously with the ethanol.

The cyclohexane results are again favourable. Inlet levels of 0.1 ppmV are approximately 5 times ISBT specifications and outlet levels are all below the measurement detection level at 0.001 ppmV, a reduction in concentration of at least 100 times. More than 1.5 bottles of contaminated CO₂ was processed during the test with the equivalent of a further bottle of clean gas with no contaminant desorption in evidence. The summary table presents the results which are also shown in the graphs. There is not a specified maximum level for Ethanol as a species, but it forms part of the volatile hydrocarbon class at 20 ppmV. All of the graphs show variations in the measured inlet contaminant concentrations. This could be due, for example, to small variations in the gas flows forming the contaminant dilution and its effect on the resulting concentration. Additionally, the low contaminant concentration levels seen throughout the work should be borne in mind, with many results in the parts per billion range. In several cases, the outlet concentration was below the limits of detection of the analytical method for that contaminant; in these cases the concentration reported is the detection level.

Conclusions :-
A device suitable for protection of the gaseous CO₂ at a licenced premises has been designed and tested by domnick hunter. Using a number of representative potential contaminants, the ‘safeguard’ protection unit has been shown to bring out of specification CO₂ back within specified limits. As such it has proved its capability in the ‘quality incident protection’ role for which it was intended. Tests have shown that residual adsorption capacity remains in the unit as no evidence of contaminant breakthrough was seen. This provides a useful safety factor, given the un-quantified nature of potential quality excursions occurring during the unit’s working lifetime, in a real application. Further development work is underway to extend the contaminant removal data. A patent application for the product is being progressed with plans to make the unit suitable for volume production and introduce it commercially.

References:-
1 R Fielding & S Kelly, (2001) 'Developments in Carbon Dioxide Purification for the Beverage Industry', ISBT Quality Control Technical Sub Committee For CO₂ Specification. Presented at BevTech 2001, Ft. Lauderdale. 2 International Society of Beverage Technologists (2001) ‘CO₂ Quality Guidelines and Analytical Procedure Bibliography’ 2nd Edition, Homosassa, FL. 3 European Industrial Gas Association, (1999), ‘Carbon Dioxide Source Certification, Quality Standards And Verification’, publication IGC Doc 70/99/E, Brussels, Belgium. 4 Compressed Gas Association (1995) ‘Carbon Dioxide Cylinder Filling and Handling Procedures’, publication CGA G-6.3, Second Edition, Chantilly, Va.