- Email: [email protected]
An industrial user's perspective on agglomeration development
Powder Technology 130 (2003) 14 – 17 www.elsevier.com/locate/powtec
An industrial user’s perspective on agglomeration development D.W. York* Proctor and Gamble, Technical Centre Ltd., P.O. Box Forest Hall No. 2, Whitley Road, Long Benton, Newcastle upon Tyne NE12 9TS, UK
Abstract Over the last decade considerable research has been published on agglomeration and associated techniques for characterising and modelling such systems. We are getting close to a fuller understanding of the process that could transform our capabilities. This article reviews the progress and suggest areas that are currently missing including characterisation, scale up and new equipment design to make the most of there learning’s. D 2002 Published by Elsevier Science B.V. Keywords: Industrial; Perspective; Agglomeration
1. Introduction I am particularly pleased to be able to give a personal viewpoint of what I have seen as published research in the area of agglomeration from an industrial user’s perspective. Before I start, you may be asking what interest is agglomeration to P&G and on what basis am I giving this talk? Firstly, P&G is a multinational consumer goods company making a range of products from Pampers to Pringles, Ariel to Oil of Olaz, Camay to Crest, Tide to Tampax. In doing so, we handle lots of powders, some which we want to prevent from agglomerating, and some we want to agglomerate— such as instant coffee, Metamucil and probably the largest volume in laundry detergents, where, as the largest producer by far of detergents in the world, we make a considerable number of particles. The major growth in agglomeration in the detergent industry came about with the move to compact products, introduced by Kao in Japan, followed by intensive work by both Unilever and us. This is because mechanically agglomerated processes produce much higher density particles than traditional spray-dried ones, allowing, along with more weight efficient chemistry, a much lower dosage volume. It is also a lower energy process and Unilever actually won a Dutch environmental award for their process. As a company, our competitive strength is in making superior products, finding ways of incorporating our latest *
Tel.: +44-1-91-279-26-76 (direct), +44-1-91-279-27-82 (secretary); fax: +44-01-91-279-27-57. E-mail address: [email protected] (D.W. York). 0032-5910/02/$ - see front matter D 2002 Published by Elsevier Science B.V. PII: S 0 0 3 2 - 5 9 1 0 ( 0 2 ) 0 0 2 1 9 - X
chemicals, often liquids, into a stable granular form. Given that we need to make granular products suited to the raw materials, and not the other way round, the importance of understanding the mechanical properties of the wet mass formed by our surfactants and powders is key to a viable agglomeration process. I have had the pleasure of developing a wide range of agglomeration processes, from bench top experiments through pilot plants to full-scale designing the production units and starting them up in Europe, as well as supporting the design and startups of our North American units. I have also had the dubious pleasure of sorting them out when they did not work as planned, so I have a wealth of practical experience to compliment the theoretical knowledge. P&G have used a range of agglomerators, both air and mechanical fluidised beds, using both hot melt and aqueous solution binder systems. In some of our processes, it is the binder material that we really want, so our focus has been to avoid agglomeration as long as possible, in order to be able to maximise the binder level, and hence active content. This brings a particular problem area, which I will address later. Finally, we have worked hard at scaling down our agglomeration processes, in order to do research and development for new materials and processes in an affordable and rapid manner. This, in itself, has been a troubled road, but if a supplier can only give you 100 g of a new material, then very soon you are forced into scaling down your work. I do not claim to be an expert, especially in the theoretical side. What I would like to do is stimulate thoughts as to where I see we are and if are we going in the right directions for the research to be of major value to industry.
D.W. York / Powder Technology 130 (2003) 14–17
I would like to cover four distinct areas: mechanisms, equipment design and scale up/down, operational perspectives, and quality. Covering this all cannot be done in depth but I hope it will provoke more detailed discussions.
2. Mechanisms As practitioners, we need to imagine what it is like for the particles in these processes in order to be able to control the process. Nearly all agglomeration processes are carried out in opaque stainless steel units so it has not been easy to see what is going on inside. In addition, as Dr. Williams of Bradford used to say, ‘‘Particles can’t read’’ so just writing the word agglomeration on a piece of equipment will not necessarily ensure that this is the process that will happen inside. To date, we have seen a lot of interpretation of the mechanisms going on inside, with Ennis et al. [1] developing a classical model of inter particle interactions, especially taking into account that the particles are not hard, inelastic balls. I have heard a lot of criticisms of the Ennis model, but to me, it has been a useful advancement in our understanding from the old Rumpf model. I question whether we will ever have the ultimate model, but like all science, an ever-nearer approximation to the truth with time. Using this interpretation, a number of papers have been written on the use of population balance modelling. One problem of these is they are often a long way from being easy to use, requiring estimation of a number of parameters, or the simple ones miss some of the key transformations, leading to inaccuracies. One interesting practical approach has been that of Hounslow, who has developed a comparison of the agglomeration process to chemical reaction kinetics as a way of understanding the controlling parameters of a melt binder agglomeration. Whilst this approach may upset the purists, it is ideal for the industrial practitioner, since it gives him a picture of what is going on and some knobs to control the process. It does have its dangers if we only lump the whole process into one. We still need to develop our understanding of the individual transformation stages, from wetting to consolidation to attrition, to develop a comprehensive model. All the models so far depend on how well two particles will stick to each other when they collide, with experimental results being used to generate these parameters. We are still a long way off being able to predict this behaviour, but some good progress has been made: first of all, with the work started by Nesbit and Conway-Jones, back in 1958, and has been further refined by Kristensen in a series of papers in 1995. These developed the concept of there being a minimum binder level needed to saturate the powders, before the binder becomes available for providing stickiness, as well as the deformability of the wetted particles being a good predictor to whether the particles would agglomerate. The next stage has been to look at processes
15
where the binder may not have sufficient time to be absorbed before the particles collide, especially with some of the viscous binders we use in the detergent industry. We therefore await the work of Prof. Litster and associates in this area. Another important area is to understand how the particles will be brought together in order to help predict the level of agglomeration, since the process is not completely random because of the geometry of the agglomerator’s equipment. Here the use of PEPT has been shown to be a very valuable research tool, in seeing the individual particle trajectories inside mixers and agglomerators. Prof. Bridgewater and Bruno Laurent have done key research in Cambridge (showing some interesting dynamics for the ploughshare type agglomerators) as well as Prof. Seville and co workers at Birmingham—incidentally using the same camera! I feel that there is a lot more information to be usefully gained by this technique, not least in being able to calibrate mathematical models as well as in aiding better equipment design. All this is helping the industrialist get a much better picture of what is happening inside the equipment, even if we cannot predict to anywhere near the same accuracy as liquid systems. The following slide, borrowed from Jim Michaels [2], shows the areas of difference between our knowledge of liquid systems and that of particulate ones.
Scale Macro
Fluid-phase processes
Distributed UO models . Reactors (CSTR, PFR) . Distillation (McCabe-Thiele) Meso Phenomenological models . Thermodynamics (activity coefficients) . Reactions (r = f(C, T)) . Transport (viscosity, mass/heat transfer coefficients) Micro and Molecular models for below prediction of phenomenological constants
Particle processes Non-distributed UO models . Population balances
?
Interparticle forces Particle Morphology
From a personal point of view, I would like to see more mechanistic understanding to build from the ground up than a lot of mathematical models at too early a stage. By a better understanding of the basic mechanisms happening, I see a way of pulling together all the knowledge we have on agglomeration into a universal model that can be used for any material or equipment.
16
D.W. York / Powder Technology 130 (2003) 14–17
3. Scale up and equipment Whilst there has been considerable progress in the understanding of the mechanisms of agglomeration, the area of equipment still seems to be lagging behind in a number of important ways. Roget describes agglomeration as a group of things gathered together haphazardly, and this is what it often seems to be: random collisions resulting a some particle growth. There are a wide variety of different styles of agglomerator designs available, but they tend to fall into a few categories: rotating drums (inclined vertically or horizontally), fluidised beds and mechanically agitated fluid beds (on a vertical or horizontal axis). However, a recent visit to ACHEMA failed to show any novel design features, only minor modifications to the existing designs and considerable copying, especially in the pharmacy area. We still see V blender style equipment sold as ideal agglomerators when it is only the cohesiveness of powders that prevent them being ideal segregators. There seems to be little attention to using the understanding of agglomeration to equipment design. Indeed, the designs seem to be fundamentally the same as 40 years ago when they were developed as powder mixers. The main improvements seem to be limited to instrumentation or better motors. The exception seems to be LoUdige with their work on different blades inside their ploughshare mixers. In essence, any system where a bed of powder is agitated can become an agglomerator if the cohesiveness of the system is high enough. Even grinders can act as agglomerators as I have found to my cost. Perhaps this is why most agglomerators started out as designed as mixers. However, the key requirements for a mixer are not the same as for ideal agglomerators. In addition, research seems to be focussed in understanding what is going on in these nonideal agglomerators, rather than in designing ones, which will give a more efficient process. For example, how to control to a narrow particle size distribution. After all, we have the fundamentals; segregation in rotating pan agglomerators has been shown to work for decades now. (In this, it is worth pointing out that the capital investment for separating non-acceptable particle sizes can be significantly greater than the capital investment in the agglomeration equipment, not to mention that with recycle, a lot of capacity can be used up providing space for the non-optimum product.) The second area of opportunity is in scale up. This is traditionally a problem for practitioners in powder systems, but doubly so for agglomerators. Of the mechanically assisted fluid bed systems, there is not a coherent criterion for scale up apart from the motor. Length to diameter ratios change, clearances are not consistent. Thus, one is always left with considerable fine-tuning in moving from one size to another. I understand that we have a lot of existing agglomerators around the world and that we would benefit from understanding more about how best to operate them. In addition,
that replacing them with more scaleable, predictable units will be costly, but that should not prevent new systems being developed for future plants. I feel the current designs are wasteful. The third area of opportunity is in scale down. With the desire to develop products faster and cheaper, it is becoming more important to be able to start development on the bench scale, especially when one is concerned with materials that are either very expensive or have hygiene concerns. Yet I am not aware of any scaled down replicas of well-known units such as Schugi, Eirich or LoUdige. Universities and companies are left to develop their own, with the consequent lost time, cost and lack of standardisation that inhibits comparison of results. Just think of the benefits to a company if they could come up with a kilo per hour unit, which gave reproducible granular characteristics to a full-scale unit, and allowed one to learn about the impact of process changes. Provide each particle technology centre with one free and you would get lots of publicity as well as some leading edge research done on your specific design! Personally I would like to see more research done into designing equipment to produce the particles properties we want, rather than all the efforts put into understanding how to get the best out of the equipment we have got. This could start by designing microscaled apparatus for understanding the transformations that raw materials undergo and measuring their properties to predict how they will behave, in much the same way we now measure fluid rheology to predict flows in pipes and mixers for liquids.
4. Operational matters This area is more like cookery with a lot of skilful operators aided by experience and a little science. That is not to say there is anything wrong in cookery as in when performed in some of the local restaurants! The best combine creativity with a good understanding of their raw materials and their process equipment. In addition, with better the understanding of the science, then the more chefs can produce acceptable results. Running an agglomeration process requires coping with wall build-up. At one level, a coating of equipment can have a big effect on how fast materials agglomerate. At another level, continuing build-up will make the process vary over time, hence difficult to control. Finally, too much will effect process reliability. In agglomeration, we are trying to put in binders to make particles sticky yet we want them not to stick to the equipment, or at least not too much. I have yet to see any studies done in this area, though there are a number of patents published in the area of using dusting as a way of controlling this effect. So what is critical to build-up and how do we control it? Associated with this is the question of clearance between the wall and the tools.
D.W. York / Powder Technology 130 (2003) 14–17
I have already mentioned that we get too wide a particle size distribution from our units with the result that we end up having to remove fines and coarse particles and recycle them. Thus, running an agglomeration process includes the important step of balancing the generation and usage of recycled materials. This is especially difficult when the porosity and surface area of the powders is not consistent and you do not want the binder level to change when this is the active ingredient. Process control strategies are very important here. In addition, the behaviour of the wet mass of powder and binder can also vary significantly with different ambient conditions. Raw material temperature can change from one delivery to the next and, since no process is sealed from the atmosphere, the humidity of the incoming air can also affect the wet mass properties to such an extent as to send a process out of control if one is not careful. This becomes critical if you are developing a process in northern England yet trying to run it in the southern United States. In addition, the absorbency properties of powders can vary from one delivery to another. Yet how does one measure this to have better predictability? Moving to automatic control brings on the additional demand of on-line measurement of the behaviour of the system. Here we have seen a lot of developments in particle size measurement. De Silva [3], in his recent article, says that almost everybody agrees that laser diffraction is the way to go. However, as Mort [4] points out in his presentation for dense phase systems, digital imaging using video cameras has a lot to offer. All this still requires a decent model of the system to plug the data into. I also wonder how the use of the developing tomographic techniques by such proponents as Prof. Williams at Leeds will also help in the area of on-line process monitoring. In addition, there is scope for using the agitators as sensors themselves, using the torque measurement to detect changes in the rheological properties of the wet mass inside, and hence the agglomeration profile.
17
controlling the energy input into these systems is one way of controlling density, especially so in mechanically agitated systems. However, this also tends to increase agglomeration levels. And what about other critical parameters such as porosity and shape? If agglomeration is not going to be seen as an old process, we will need also to learn how to effectively determine these parameters as well. A small point here is that there is still a large discussion as to how to measure a number of key parameters such as flow properties, porosity, etc. Also, a lack of agreement here will slow down the ability of the industry to move forward in our understanding.
6. Conclusion What I have tried to do is to paint a personal picture of the key areas of need in future research in agglomeration. A lot of progress has been made but there are lots of open spaces ripe for industrially relevant discoveries and consequently attract industrial funds! Yet where to put the effort? De Silva, in his recent article on Characterisation of Particulate Materials, bemoans the fact that, whilst there is a huge need for work in over 20 different characteristics, most of the work is done on particle size measurement. I can sympathise. As a practitioner, I am a great supporter of the Pareto principle; that is, 80% of the information can be obtained with 20% of the effort and I would rather see research and development spread out over the whole of the field of agglomeration than concentrate on a few fashionable areas. In particular, we seem to have a plethora of mathematical modellers and too few experimenters. If nothing else, I hope this paper has at least stimulated some debate on the subject.
References 5. Quality and other matters Agglomeration is primarily about enlarging the particle size of the starting powders. However, as we learn more, there is a realisation that other properties are important to the performance of the particles. One of these is the particle density. As Mort describes, we have learnt that
[1] B.J. Ennis, G.I. Tardos, R. Pfeffer, Powder Technology 65 (1991) 257 – 272. [2] J.N. Michaels, Research needs in particle science and technology, AIChE, 1999. [3] S.R. de Silva, Characterisation of particulate materials, Powder Handling and Processing 12 (4) (2000) 355 – 363. [4] P.R. Mort, Control of agglomerate product attributes, Powder Technol. 117 (1 – 2) (2001) 173 – 176.
An industrial user’s perspective on agglomeration development D.W. York* Proctor and Gamble, Technical Centre Ltd., P.O. Box Forest Hall No. 2, Whitley Road, Long Benton, Newcastle upon Tyne NE12 9TS, UK
Abstract Over the last decade considerable research has been published on agglomeration and associated techniques for characterising and modelling such systems. We are getting close to a fuller understanding of the process that could transform our capabilities. This article reviews the progress and suggest areas that are currently missing including characterisation, scale up and new equipment design to make the most of there learning’s. D 2002 Published by Elsevier Science B.V. Keywords: Industrial; Perspective; Agglomeration
1. Introduction I am particularly pleased to be able to give a personal viewpoint of what I have seen as published research in the area of agglomeration from an industrial user’s perspective. Before I start, you may be asking what interest is agglomeration to P&G and on what basis am I giving this talk? Firstly, P&G is a multinational consumer goods company making a range of products from Pampers to Pringles, Ariel to Oil of Olaz, Camay to Crest, Tide to Tampax. In doing so, we handle lots of powders, some which we want to prevent from agglomerating, and some we want to agglomerate— such as instant coffee, Metamucil and probably the largest volume in laundry detergents, where, as the largest producer by far of detergents in the world, we make a considerable number of particles. The major growth in agglomeration in the detergent industry came about with the move to compact products, introduced by Kao in Japan, followed by intensive work by both Unilever and us. This is because mechanically agglomerated processes produce much higher density particles than traditional spray-dried ones, allowing, along with more weight efficient chemistry, a much lower dosage volume. It is also a lower energy process and Unilever actually won a Dutch environmental award for their process. As a company, our competitive strength is in making superior products, finding ways of incorporating our latest *
Tel.: +44-1-91-279-26-76 (direct), +44-1-91-279-27-82 (secretary); fax: +44-01-91-279-27-57. E-mail address: [email protected] (D.W. York). 0032-5910/02/$ - see front matter D 2002 Published by Elsevier Science B.V. PII: S 0 0 3 2 - 5 9 1 0 ( 0 2 ) 0 0 2 1 9 - X
chemicals, often liquids, into a stable granular form. Given that we need to make granular products suited to the raw materials, and not the other way round, the importance of understanding the mechanical properties of the wet mass formed by our surfactants and powders is key to a viable agglomeration process. I have had the pleasure of developing a wide range of agglomeration processes, from bench top experiments through pilot plants to full-scale designing the production units and starting them up in Europe, as well as supporting the design and startups of our North American units. I have also had the dubious pleasure of sorting them out when they did not work as planned, so I have a wealth of practical experience to compliment the theoretical knowledge. P&G have used a range of agglomerators, both air and mechanical fluidised beds, using both hot melt and aqueous solution binder systems. In some of our processes, it is the binder material that we really want, so our focus has been to avoid agglomeration as long as possible, in order to be able to maximise the binder level, and hence active content. This brings a particular problem area, which I will address later. Finally, we have worked hard at scaling down our agglomeration processes, in order to do research and development for new materials and processes in an affordable and rapid manner. This, in itself, has been a troubled road, but if a supplier can only give you 100 g of a new material, then very soon you are forced into scaling down your work. I do not claim to be an expert, especially in the theoretical side. What I would like to do is stimulate thoughts as to where I see we are and if are we going in the right directions for the research to be of major value to industry.
D.W. York / Powder Technology 130 (2003) 14–17
I would like to cover four distinct areas: mechanisms, equipment design and scale up/down, operational perspectives, and quality. Covering this all cannot be done in depth but I hope it will provoke more detailed discussions.
2. Mechanisms As practitioners, we need to imagine what it is like for the particles in these processes in order to be able to control the process. Nearly all agglomeration processes are carried out in opaque stainless steel units so it has not been easy to see what is going on inside. In addition, as Dr. Williams of Bradford used to say, ‘‘Particles can’t read’’ so just writing the word agglomeration on a piece of equipment will not necessarily ensure that this is the process that will happen inside. To date, we have seen a lot of interpretation of the mechanisms going on inside, with Ennis et al. [1] developing a classical model of inter particle interactions, especially taking into account that the particles are not hard, inelastic balls. I have heard a lot of criticisms of the Ennis model, but to me, it has been a useful advancement in our understanding from the old Rumpf model. I question whether we will ever have the ultimate model, but like all science, an ever-nearer approximation to the truth with time. Using this interpretation, a number of papers have been written on the use of population balance modelling. One problem of these is they are often a long way from being easy to use, requiring estimation of a number of parameters, or the simple ones miss some of the key transformations, leading to inaccuracies. One interesting practical approach has been that of Hounslow, who has developed a comparison of the agglomeration process to chemical reaction kinetics as a way of understanding the controlling parameters of a melt binder agglomeration. Whilst this approach may upset the purists, it is ideal for the industrial practitioner, since it gives him a picture of what is going on and some knobs to control the process. It does have its dangers if we only lump the whole process into one. We still need to develop our understanding of the individual transformation stages, from wetting to consolidation to attrition, to develop a comprehensive model. All the models so far depend on how well two particles will stick to each other when they collide, with experimental results being used to generate these parameters. We are still a long way off being able to predict this behaviour, but some good progress has been made: first of all, with the work started by Nesbit and Conway-Jones, back in 1958, and has been further refined by Kristensen in a series of papers in 1995. These developed the concept of there being a minimum binder level needed to saturate the powders, before the binder becomes available for providing stickiness, as well as the deformability of the wetted particles being a good predictor to whether the particles would agglomerate. The next stage has been to look at processes
15
where the binder may not have sufficient time to be absorbed before the particles collide, especially with some of the viscous binders we use in the detergent industry. We therefore await the work of Prof. Litster and associates in this area. Another important area is to understand how the particles will be brought together in order to help predict the level of agglomeration, since the process is not completely random because of the geometry of the agglomerator’s equipment. Here the use of PEPT has been shown to be a very valuable research tool, in seeing the individual particle trajectories inside mixers and agglomerators. Prof. Bridgewater and Bruno Laurent have done key research in Cambridge (showing some interesting dynamics for the ploughshare type agglomerators) as well as Prof. Seville and co workers at Birmingham—incidentally using the same camera! I feel that there is a lot more information to be usefully gained by this technique, not least in being able to calibrate mathematical models as well as in aiding better equipment design. All this is helping the industrialist get a much better picture of what is happening inside the equipment, even if we cannot predict to anywhere near the same accuracy as liquid systems. The following slide, borrowed from Jim Michaels [2], shows the areas of difference between our knowledge of liquid systems and that of particulate ones.
Scale Macro
Fluid-phase processes
Distributed UO models . Reactors (CSTR, PFR) . Distillation (McCabe-Thiele) Meso Phenomenological models . Thermodynamics (activity coefficients) . Reactions (r = f(C, T)) . Transport (viscosity, mass/heat transfer coefficients) Micro and Molecular models for below prediction of phenomenological constants
Particle processes Non-distributed UO models . Population balances
?
Interparticle forces Particle Morphology
From a personal point of view, I would like to see more mechanistic understanding to build from the ground up than a lot of mathematical models at too early a stage. By a better understanding of the basic mechanisms happening, I see a way of pulling together all the knowledge we have on agglomeration into a universal model that can be used for any material or equipment.
16
D.W. York / Powder Technology 130 (2003) 14–17
3. Scale up and equipment Whilst there has been considerable progress in the understanding of the mechanisms of agglomeration, the area of equipment still seems to be lagging behind in a number of important ways. Roget describes agglomeration as a group of things gathered together haphazardly, and this is what it often seems to be: random collisions resulting a some particle growth. There are a wide variety of different styles of agglomerator designs available, but they tend to fall into a few categories: rotating drums (inclined vertically or horizontally), fluidised beds and mechanically agitated fluid beds (on a vertical or horizontal axis). However, a recent visit to ACHEMA failed to show any novel design features, only minor modifications to the existing designs and considerable copying, especially in the pharmacy area. We still see V blender style equipment sold as ideal agglomerators when it is only the cohesiveness of powders that prevent them being ideal segregators. There seems to be little attention to using the understanding of agglomeration to equipment design. Indeed, the designs seem to be fundamentally the same as 40 years ago when they were developed as powder mixers. The main improvements seem to be limited to instrumentation or better motors. The exception seems to be LoUdige with their work on different blades inside their ploughshare mixers. In essence, any system where a bed of powder is agitated can become an agglomerator if the cohesiveness of the system is high enough. Even grinders can act as agglomerators as I have found to my cost. Perhaps this is why most agglomerators started out as designed as mixers. However, the key requirements for a mixer are not the same as for ideal agglomerators. In addition, research seems to be focussed in understanding what is going on in these nonideal agglomerators, rather than in designing ones, which will give a more efficient process. For example, how to control to a narrow particle size distribution. After all, we have the fundamentals; segregation in rotating pan agglomerators has been shown to work for decades now. (In this, it is worth pointing out that the capital investment for separating non-acceptable particle sizes can be significantly greater than the capital investment in the agglomeration equipment, not to mention that with recycle, a lot of capacity can be used up providing space for the non-optimum product.) The second area of opportunity is in scale up. This is traditionally a problem for practitioners in powder systems, but doubly so for agglomerators. Of the mechanically assisted fluid bed systems, there is not a coherent criterion for scale up apart from the motor. Length to diameter ratios change, clearances are not consistent. Thus, one is always left with considerable fine-tuning in moving from one size to another. I understand that we have a lot of existing agglomerators around the world and that we would benefit from understanding more about how best to operate them. In addition,
that replacing them with more scaleable, predictable units will be costly, but that should not prevent new systems being developed for future plants. I feel the current designs are wasteful. The third area of opportunity is in scale down. With the desire to develop products faster and cheaper, it is becoming more important to be able to start development on the bench scale, especially when one is concerned with materials that are either very expensive or have hygiene concerns. Yet I am not aware of any scaled down replicas of well-known units such as Schugi, Eirich or LoUdige. Universities and companies are left to develop their own, with the consequent lost time, cost and lack of standardisation that inhibits comparison of results. Just think of the benefits to a company if they could come up with a kilo per hour unit, which gave reproducible granular characteristics to a full-scale unit, and allowed one to learn about the impact of process changes. Provide each particle technology centre with one free and you would get lots of publicity as well as some leading edge research done on your specific design! Personally I would like to see more research done into designing equipment to produce the particles properties we want, rather than all the efforts put into understanding how to get the best out of the equipment we have got. This could start by designing microscaled apparatus for understanding the transformations that raw materials undergo and measuring their properties to predict how they will behave, in much the same way we now measure fluid rheology to predict flows in pipes and mixers for liquids.
4. Operational matters This area is more like cookery with a lot of skilful operators aided by experience and a little science. That is not to say there is anything wrong in cookery as in when performed in some of the local restaurants! The best combine creativity with a good understanding of their raw materials and their process equipment. In addition, with better the understanding of the science, then the more chefs can produce acceptable results. Running an agglomeration process requires coping with wall build-up. At one level, a coating of equipment can have a big effect on how fast materials agglomerate. At another level, continuing build-up will make the process vary over time, hence difficult to control. Finally, too much will effect process reliability. In agglomeration, we are trying to put in binders to make particles sticky yet we want them not to stick to the equipment, or at least not too much. I have yet to see any studies done in this area, though there are a number of patents published in the area of using dusting as a way of controlling this effect. So what is critical to build-up and how do we control it? Associated with this is the question of clearance between the wall and the tools.
D.W. York / Powder Technology 130 (2003) 14–17
I have already mentioned that we get too wide a particle size distribution from our units with the result that we end up having to remove fines and coarse particles and recycle them. Thus, running an agglomeration process includes the important step of balancing the generation and usage of recycled materials. This is especially difficult when the porosity and surface area of the powders is not consistent and you do not want the binder level to change when this is the active ingredient. Process control strategies are very important here. In addition, the behaviour of the wet mass of powder and binder can also vary significantly with different ambient conditions. Raw material temperature can change from one delivery to the next and, since no process is sealed from the atmosphere, the humidity of the incoming air can also affect the wet mass properties to such an extent as to send a process out of control if one is not careful. This becomes critical if you are developing a process in northern England yet trying to run it in the southern United States. In addition, the absorbency properties of powders can vary from one delivery to another. Yet how does one measure this to have better predictability? Moving to automatic control brings on the additional demand of on-line measurement of the behaviour of the system. Here we have seen a lot of developments in particle size measurement. De Silva [3], in his recent article, says that almost everybody agrees that laser diffraction is the way to go. However, as Mort [4] points out in his presentation for dense phase systems, digital imaging using video cameras has a lot to offer. All this still requires a decent model of the system to plug the data into. I also wonder how the use of the developing tomographic techniques by such proponents as Prof. Williams at Leeds will also help in the area of on-line process monitoring. In addition, there is scope for using the agitators as sensors themselves, using the torque measurement to detect changes in the rheological properties of the wet mass inside, and hence the agglomeration profile.
17
controlling the energy input into these systems is one way of controlling density, especially so in mechanically agitated systems. However, this also tends to increase agglomeration levels. And what about other critical parameters such as porosity and shape? If agglomeration is not going to be seen as an old process, we will need also to learn how to effectively determine these parameters as well. A small point here is that there is still a large discussion as to how to measure a number of key parameters such as flow properties, porosity, etc. Also, a lack of agreement here will slow down the ability of the industry to move forward in our understanding.
6. Conclusion What I have tried to do is to paint a personal picture of the key areas of need in future research in agglomeration. A lot of progress has been made but there are lots of open spaces ripe for industrially relevant discoveries and consequently attract industrial funds! Yet where to put the effort? De Silva, in his recent article on Characterisation of Particulate Materials, bemoans the fact that, whilst there is a huge need for work in over 20 different characteristics, most of the work is done on particle size measurement. I can sympathise. As a practitioner, I am a great supporter of the Pareto principle; that is, 80% of the information can be obtained with 20% of the effort and I would rather see research and development spread out over the whole of the field of agglomeration than concentrate on a few fashionable areas. In particular, we seem to have a plethora of mathematical modellers and too few experimenters. If nothing else, I hope this paper has at least stimulated some debate on the subject.
References 5. Quality and other matters Agglomeration is primarily about enlarging the particle size of the starting powders. However, as we learn more, there is a realisation that other properties are important to the performance of the particles. One of these is the particle density. As Mort describes, we have learnt that
[1] B.J. Ennis, G.I. Tardos, R. Pfeffer, Powder Technology 65 (1991) 257 – 272. [2] J.N. Michaels, Research needs in particle science and technology, AIChE, 1999. [3] S.R. de Silva, Characterisation of particulate materials, Powder Handling and Processing 12 (4) (2000) 355 – 363. [4] P.R. Mort, Control of agglomerate product attributes, Powder Technol. 117 (1 – 2) (2001) 173 – 176.