Revealing the secrets of fluidised beds: Exploiting links between academia and industry


Exploiting links between academia and industry

Close cooperation between academics and industrialists over many years can be of great benefit to both parties. David Newton describes how this has led to the development of a unique X-ray imaging technique to visualise and characterise flows in multiphase systems.


Occasionally I am asked, by both academics and industrialists, whether long-term collaboration with academia can pay off. Having worked as a chemical engineer in industry for the past 20 years, I have developed close contacts with several UK university departments and can, from an industrialist’s perspective, answer a resounding ‘yes’ to that question. Although it may take several years to fully realise such a ‘pay-off’, it is well worth the time, expense and sometimes trouble for industry to develop and maintain such links with academia.

The purpose of this article is to highlight one such long-term relationship and outline how it has transformed the way some basic chemical engineering practices are now carried out in BP. This is very much a personal account of the development of a unique X-ray imaging technique to visualise and characterise flows in gas–solid and gas–liquid–solid multiphase systems. This technique has contributed much to the technical understanding of reactor multiphase flows and to commercial successes in several areas, and has also generated significant financial rewards.

Three frequently asked questions concerning X-ray imaging are ‘Why?’ as in ‘Why do we use the technique?’ ‘How?’ as in ‘How does it work and how did BP come to use and develop it?’ and ‘What?’ as in ‘What are the technical and financial benefits?’ In this article I intend to answer these questions in some detail. But I would also like to address the questions ‘Where?’ and ‘Who?’, as in ‘Where did the technique come from?’ and ‘Who was responsible for initiating and applying the technique?’

Historical development – who and where?

It is over 40 years since Peter Rowe and colleagues, working at the Atomic Energy Research Establishment (AERE) Harwell, and then University College London (UCL), first demonstrated the immense potential of applying X-ray imaging to visualise the internal flow patterns of bubbles in gas fluidised beds. Since those early pioneering days there have been many hundreds of publications dealing with the properties of gas fluidised beds obtained from the X-ray imaging technique,1 a significant proportion of them coming from UCL as scientific papers or PhD theses. In 1965 Peter Rowe left AERE and moved to UCL as Ramsay Head and Professor of Chemical Engineering. Since then the technique has flourished over the years at UCL, mainly through the efforts of Peter Rowe (who retired in 1985), John Yates (current Head of Chemical Engineering at UCL), David Cheesman and, more recently, Paola Lettieri.

Back in 1985 BP Research began what was to be a long and very fruitful relationship with the Department of Chemical Engineering at UCL. Over the following seven years BP (and others, such as The Dupont Chemical Co.) used and helped develop the facility at UCL. At that time, we were driven by the need to understand what was going on inside some of our gas fluidised bed processors. We had begun to doubt some of the design bases that had been used for designing commercial scale fluidised bed process hardware. During 1985 to 1991, whilst working at UCL, we did find many discrepancies between phenomena and data determined from the X-ray technique and those conforming to an ‘established view’. Many of these findings have been published.2–4

How does it work?

Differences in X-ray absorbance of the components (i.e. solids, liquids or gases) present in a reactor vessel allow hydrodynamic information to be determined. X-rays do not interact with the hydrodynamic processes and therefore do not disturb the measurements, in contrast to techniques that use intrusive probes. In the X-ray technique a pulsed (typically 50 Hz) high energy beam (from 40 to 150 kV) is produced from a rotating anode. The X-ray pulses (typically between 1 and 5 ms in duration for rapidly fluctuating multiphase systems) are synchronised with an image capturing device and pass through the reactor vessel, where X-ray absorption is proportional to the nature and quantity of material along the path. Rapid X-ray pulses of a few milliseconds duration are essential for accurate quantitative work.

The X-ray beam emerging from the reactor vessel is amplified using an image intensifier, without which it would be impossible to detect the small changes of intensity in the emerging X-ray beam. The image intensifier tube converts the X-ray absorption patterns into a light image of sufficient brightness and contrast to be recorded by some means, typically photographic film (cine camera), charge coupled device or video camera. Figure 1 shows a schematic of the X-ray system and Figure 2 shows a schematic of the BP recirculating loop configuration for imaging a 750 mm diameter fluidised bed unit.

Why use the X-ray technique?

Why use X-rays? Why visualise multiphase systems? If this article were concerned with medical applications then the answer to the first question would be self-evident. It is almost inconceivable, certainly in the Western world, to confirm or diagnose many medical conditions, from broken bones to lung tumours, without some form of X-ray imaging technology.

In chemical engineering, as in many industrial processes, a technique that enables characterisation, and provides detailed information, of flow patterns inside reactor vessels or transfer lines would be highly desirable and extremely beneficial. Eminent researchers have highlighted the problems encountered by chemical engineers in predicting multiphase flow behaviour. Derek Geldart5 considered that ‘the arrival time of a space probe traveling to Saturn can be predicted more accurately than the behaviour of a fluidised bed reactor’. And G. F. Hewitt6 stated: ‘multiphase flows are often extremely complex in nature and many of the relationships used for multiphase flows are of an essentially empirical nature, are of limited applicability, and reflect the poor physical understanding of many two-phase flow phenomena’.

X-ray imaging can be used both as a diagnostic tool – to help identify the cause of the ‘problem’ and also to help formulate the ‘solution’ to that problem – and as a way of understanding multiphase flow phenomena and their relevance to commercial operational parameters. In the context of chemical engineering, the images are usually taken from reactor mock-ups, pilot plants or pieces of process equipment operating at, or as close as possible to, realistic process conditions. This can mean performing experiments at elevated temperatures and pressures, with real-scale process hardware and catalyst systems.

Although the theoretical aspects of many multiphase flow phenomena are described by empirical correlations, it is clear that the application of many of these correlations for industrial exploitation can be somewhat uncertain. We found such uncertainties during early work on the X-ray facility at UCL in the 1980s. The object of the investigation was to test the jet penetration lengths from horizontal and vertical orientated nozzles for use in a fluid catalytic cracking (FCC) regenerator unit.

The objective of the FCC process is to use both catalytic and thermal cracking to convert heavy hydrocarbons (high molecular weight) into lower molecular weight compounds – usually boiling in the gasoline and middle oil ranges. The cracking reactions are endothermic and are accompanied by carbon deposition on the solid catalyst surface. The FCC process has one location for the absorption of heat, the reaction and carbon deposition, referred to as the riser, and a second location where the deposited carbon is burnt off and heat is released, referred to as the regenerator unit. The heat is then returned to the first location to drive the reaction, and the circulation of the solid catalyst particles is the means by which heat is transported around the unit. The only way this can be done efficiently is by employing one or more fluidised beds.

Many traditional modelling techniques determine the gas and solids flow properties of fluidised beds in small- or medium-scale perspex or glass equipment operating at ambient conditions, with measurements made through viewing screens or with the aid of sensors placed in the bed. Observations are not only restricted but often distorted by the presence of internal surfaces or from effects of scale-down. Consequently, most jet penetration lengths measured by such techniques fail to detect interactions between adjacent nozzles or pipes because of the stabilising effects of the viewing screens. With X-ray imaging, such limitations do not exist and unique information that cannot be obtained by other techniques is generated.

Figure 3 shows how wall effects can influence the measurements of gas jet penetration lengths in fluidised beds. Initially a nozzle was placed close to the vessel wall, made from perspex, and the jet penetration lengths were observed visually as a function of increasing nozzle exit velocity. The penetration lengths (shown by red square symbols) were hindered because of the stabilising effects of the vessel wall. The nozzle was then moved to the centre of the fluidised bed, about 150 mm away from the nearest wall, and the penetration lengths were determined by means of X-ray imaging. These penetration lengths (shown by black circle symbols) were substantially reduced because they were out of range of the stabilising influence of the vessel wall. These internal penetration lengths could be viewed only by means of X-ray imaging because of the opaqueness of the fluidised solids. Thus by use of X-ray imaging we were able to demonstrate how inaccurate and inappropriate many literature correlations are for predicting jet penetration lengths and this can have a significant financial impact on a process’s profitability.

These results are consistent with other jet penetration data determined from X-ray imaging7. As can be seen from Figure 3 agreement with the correlations is not that good.

Figure 4 shows what happens when several regenerator nozzles are arranged around the circumference of the air ring. If nozzles are placed too close to each other, as illustrated for the five-nozzle case in Figure 4a, then gas from lateral nozzles weeps around the ring to coalesce with the vertical gas nozzle to form a large plume above the central nozzle. Only with adequate nozzle separation in terms of both distance and nozzle angle, as in Figure 4b, can this region be bypassed and nozzle interactions eliminated. It is critically important for design and operational purposes to be able to predict adjacent nozzle interactions and the jet penetrations from nozzles in fluidised beds8.

Those early years at UCL became a watershed in terms of ‘opening our eyes’ to the many differences and pitfalls that exist in trying to exploit published data and correlations for industrial-scale processes without due validation. This is a lesson, I believe, we have learnt well. This may seem a rather trivial example, but I shall demonstrate how such a situation can have a very significant impact on operational costs and profitability for FCC operation.

What do we gain?

The X-ray technique has the following benefits:

  • it is a non-invasive technique that does not interfere with the system

  • an internal picture of the process fluid dynamics is generated

  • real-time information can be captured directly onto film or video

  • unambiguous images of rapidly changing multiphase flows are obtained

  • it is unaffected by process operating conditions

  • realistic process conditions can be employed for the generation of data (100 bar and 1000°C have so far been employed)

  • large-scale three-dimensional process plant can be imaged

  • it is a diagnostic tool for assessing engineering designs and ‘troubleshooting’

  • it aids flow characterisation and process optimisation

  • quantitative information about the distribution of material can be obtained.

In an FCC regenerator unit, efficient air distribution is essential to give good fluidisation of the catalyst bed. Good fluidisation is necessary to achieve stable catalyst flow into the hoppers and standpipes, as well as a fluidised bed environment into which the cyclone dip-legs discharge. Distributor defects, whether as a result of poor design or operational damage, can lead to:

  • uneven regenerator temperatures, that is afterburn

  • excessive air and power consumption

  • reductions in feed capacity/coke burning

  • poor fluidisation

  • uneven bed densities

  • unstable flow in catalyst standpipes

  • catalyst circulation problems

  • and flooding of cyclones.

The consequences of operating a regenerator unit with a damaged or improperly designed distributor can be costly. Regenerator air bypassing of the catalyst bed can lead to: 1 excessive air and power consumption – for a unit operating with 2–3% oxygen in the flue gas instead of less than 1%, the cost can be 4–5 cents per barrel 2 reduction in coke burning capacity or feed capacity – depending on the FCC processing margin, this could represent a loss of $2–3 per barrel of capacity.

If the penetration of air into the fluidised bed regenerator unit is only a fraction of that predicted by the ‘design’ correlation, due either to using an inappropriate correlation or to the nozzles being placed too close to each other around the aeration rings, resulting in nozzle–gas interactions (as in Figure 4a), then considerable quantities of air and power are required in an attempt to maintain regenerator performance. For example, for an FCC unit with 50 000 barrels per day capacity, this could have a cost penalty in excess of $1 million per year for badly designed regenerator air rings. If the design and operation of the regenerator unit was so bad that it grossly inhibited coke burning capacity or feed capacity then a cost penalty in excess of $50 million per year could be obtained. On the global scale, FCC units account for over 10 million barrels per day capacity. If, say, only 10% of these units were operating in the non-optimal way described above, then potentially this seemingly trivial design defect could account for over $1000 million per year in lost profits!

Other areas of application

In the chemicals area, the X-ra y technique has also made a significant contribution to the development of new technology, principally the injection of liquid hydrocarbons into a fluidised bed gas-phase polymerisation reactor for removal of the exothermic heat of reaction. In this technology, termed ‘High Productivity’, liquid is sprayed directly into the fluidised bed with nozzles9–10 and the increased cooling capacity achieved through the latent heat of vaporisation enables significant increases in reactor productivity. The locations and orientations of the liquid sprays were determined experimentally by visualising the vaporisation of liquid hydrocarbons in a polyethylene fluidised bed operating at realistic process conditions. Figure 5 is a photograph of the experimental equipment employed in the development of High Productivity. Previous forms of gas-phase technology that returned a mixture of gas and liquid into the base of the fluidised bed reactor through the grid plate are limited by the maximum amount of liquid that can be held in the gas stream. The limit is reached when an overload of liquid occurs. Because the process operates close to the polymer’s melting point, accurate temperature control and fluidisation characteristics are essential. The polymerisation rate and heat removal must be controlled to prevent fusion, agglomeration of polymer particles, or a thermal runaway reaction.

Liquid flooding or frothing in the bottom head of the reactor, even at relatively low levels of liquid in the recycle stream, has been reported,11 thereby restricting heat removal from the fluidised bed and causing instability. Figure 6 shows a schematic of BP’s Enhanced High Productivity loop. The heat exchanger unit is operated at a sufficiently low temperature to form liquid. The liquid is separated from the recycle fluidising gas by a separator unit and then is pumped into the fluidised bed reactor via nozzles located within the fluidised bed of growing polymer particles. Catalyst and/or catalyst modifiers can be added to the liquid for maximum dispersion within the fluidised bed. Key success factors in developing the spray nozzle technology for the polyethylene process were the ability to visualise and relate the spray characteristics from the various nozzle designs with the fluid dynamics of the polymerisation process.

In addition to polyethylene technology, the X-ray facility has been used extensively to design BP’s new fluidised bed vinyl acetate process. Process hardware such as the internal fluidised bed reactor cooling tubes, oxygen injection system and acid spray nozzles were all developed through Xray imaging.

The future

Current technology produces two-dimensional images of the internal flow patterns. As with other imaging techniques, some specialized expertise is necessary to interpret the images and make diagnoses. Extending the Xray technique to produce three-dimensional images would remove many of the uncertainties in the interpretations and could be achieved by simultaneously producing two images at right angles to each other (by using two X-ray sources/image capturing devices) and reconstructing the three-dimensional image along the same lines as a medical CAT scanner. Applying two- or three-dimensional Xray visualisation to ‘real’ plants with process chemistry occurring would further enhance the diagnostic power of the technique.

In the immediate future, by combining the X-ray technique with other complementary techniques, such as positron emission particle tracking (PEPT), a complete picture of the fluid dynamics of multiphase systems could be obtained. With X-ray imaging, individual particles cannot be seen and so no detailed information about the solids’ movement in multiphase reactors can be obtained.

In PEPT the movements of individual particles can be tracked and recorded. This technique is exploited by Birmingham University in many chemical engineering applications and has the potential to be as valuable to industry as X-ray imaging has been to BP – and it is all based on the original 1960s concept of Peter Rowe!


I hope I have demonstrated that close collaboration between academics and industrialists can pay off and not just financially. The mutual exchange of knowledge and information coupled with insights gained along the way can really foster thinking that is ‘outside the box’. For both industry and academia it can offer a way to challenge and overcome some well established but sometimes erroneous views.


  1. Davidson, J.F. and Harrison, D. (eds) (1971) Fluidisation, Academic Press, Chapter 4.
  2. Newton, D. and Johns, D.M. (1995) Fluidisation VIII, pp. 467–474.
  3. Newton, D. and Becker, S. (Summer 1996) PTQ, pp. 43–50.
  4. Newton, D., Fiorentino M. and Smith, G.B. (2001) Powder Technology 120, pp. 70–75.
  5. Geldart, D. (1986) Gas Fluidisation Technology, John Wiley & Sons Ltd.
  6. Hewitt, G.F. (1992) Handbook of Heat Exchange Design. Begell House Inc, New York.
  7. Newton, D., Smith, G.B. and Gamblin, B. (September 1997) ‘2nd European conference on fluidisation’, Bilbao, Spain.
  8. Newton, D., Grant, C.S. and Gamblin, B. (Winter 1994) HTI Quarterly, pp. 41–45.
  9. Newton, D., Chinh, J.C. and Power, M. (Autumn 1995) ‘Petrochemical and gas processing’, HTI Quarterly, pp. 81–91.
  10. Newton, D., Chinh, J.C. and Power, M. (March 1998) Hydrocarbon Processing, pp. 85–91.
  11. Union Carbide, EP 173261.

David Newton

Research Associate Fluidisation, BP Chemicals, Sunbury-on-Thames

David Newton has been responsible for developing and applying the Xray imaging technique throughout the BP group for the past 16 years. He has published over a hundred papers and patents on the applications of X-ray imaging to fluidised bed processes and multiphase flow systems. He gives both in-house and external courses in industrial fluidization in the UK and abroad.

Download Article 714KB