Applicational design and sizing of Decanter Centrifuges: Different approaches to achieve optimum performance M. H. Kopf PIERALISI Deutschland GmbH / Mannheim University of Applied Sciences, Germany
ABSTRACT More than 2.000 applications where Decanter Centrifuges are used to separate solid particles from a liquid or simultaneously separate two immiscible liquids and a solid are listed in relevant literature [1]. Of course a number of those applications are in a way exotic ones, are niche applications. Others are really volume applications, e.g. sludge thickening and dewatering, process-integrated recycling of valuable material and juice production. It is very obvious that the wide variety of applications implicate the need of significantly different Decanter Centrifuges in terms of individual features / design and size. On first sight, a Decanter Centrifuge may appear as a “simple” machine consisting of a rotating bowl and a conveyor rotating with differential speed. The “turbulent” journey, moving with the liquid flow from the inlet to the outlet will show us that a lot of thoughts must be spent to appropriately design and size this machine to reach the goal: Optimum performance in a specific application. Following the flow from entering into the feed-pipe down to the liquid outlet on one hand and the solids movement caused by the conveyor, we come across many critical aspects of state-of-the-art Decanter design and sizing e.g.: x Inlet (feed) zone design x Bowl design and sizing, x Conveyor design, x Liquid discharge design.
Figure 1: A Decanter Centrifuge, overview
1. THE INLET (FEED) ZONE This is the spot where the feed suspension is accelerated to reach bowl-speed. But it’s not as easy as it sounds. During acceleration, the feed is subjected to shear stress, thus possibly causing alterations of the particle- or droplet size distribution, i.e. particles or droplets could be reduced in size quite a lot. This is especially a problem when handling shear sensitive systems like flocks or emulsions.
The guiding principle of designing a feed zone therefore should be to minimize the shear forces acting on the feed. This gets more important the higher the speed of the centrifuge is [8].
Figure 2: The feed zone
And another aspect needs to be thought about: The possible effects of the pre-accelerated feed entering into pond. In a traditional counter-current Decanter the entering point of the feed into the pond is near to the joint of cylindrical and conical part. At this point we face pre-thickened solids / sediments being transported by the conveyor in the direction of solids discharge. The possible effect of feed entering into the pond with high radial velocity could be a re-mixing of already settled solids / sediment. Over time a lot of different geometries have been tested. The mother of all special feed zone designs is the so-called BD-Cone feed zone invented by Sharples more than 20 years ago. This design successfully implemented the principle of minimizing shear stress and avoiding remixing. The incoming feed is fed into a conical feed-cup and distributed equally in radial direction. The flow resistance of the cup results in a hold-up volume inside the cup reaching up to the feed pipe. This guarantees for smooth liquid-liquid acceleration. The pre-accelerated feed enters into the pond by means of holes equally distributed over 360°. These holes must not give an additional pressure drop to the feed. A so-called double cone takes care of the re-mixing problem because it separates the solids / sediment zone from the liquid zone. The main application of the BD-cone inlet at the days of its invention was, and is today, sludge thickening and dewatering as part of waste water treatment. Later on we will come across another feature of the BD-Technology: The “negative” operation.
Figure 3: Sharples BD-3 feed zone
Of course today there are different other geometries on the market. The design of the feed zone and the flow dynamics caused by it, is one of the secrets of the decanter producers. To help the user judging the “quality” of a feed zone, below find some basic rules: x The feed zone should have a built-in device to guide the flow and distribute it equally over 360°. x The acceleration should take place via liquid / liquid contact. x The feed flow should be guided in a way to minimize turbulences. (This is the point of “construction secrets“ of the producers)
x
The axial position of the feed pipe should be adjustable (To adapt the flow dynamics to changing feed rates)
2. BOWL SIZE AND GEOMETRY For the purpose of this paper let’s reduce the problem to three major aspects: The length and diameter of the bowl, the pond depth and the half-cone angle. The sizing, means how “big” the machine is, is usually based on pilot tests and scale-up theories, e.g. the equivalent settling area (Sigma) [2], [3], [4] Here the confusion starts. Today not less than seven different equations to calculate for Sigma are used. These last from the equations given by Sokolov [5] and Trawinski [6] to producer-adapted ones focusing on special design philosophies. The difference in all those equations mainly is: x The radius taken to calculate for the relevant G-force, x The clarification length x The percentage of the cone that is taken into account.
Figure 4: View inside a Decanter bowl
Let’s start with the “Radius” and the lengthÆ Sigma Theory Stokes’ law of settling in the centrifugal field shows that the settling velocity increases linear with the radius. This is based on the assumption that the radial settling distance for the particle is much less than it’s distance from the axis of rotation. This is almost true for shallow-pond Decanter (means low pond depth compared to the diameter) but it is highly questionable for deep-pond machines. Nevertheless the equations based on Stokes are widely used. If the radius is assumed to be of linear influence there are basically three possibilities to calculate for the acting G-Force: The pond surface radius ri (Ælowest G-Force figure), the mean radius of the pond depth rm (Æ mean G-Force figure) and the inner bowl radius re (Æ highest G-Force figure). Eventually it depends on the company’s philosophy which concept will be used. The radius and subsequently the G-Force are needed to calculate for the equivalent settling area Sigma, e.g. 6 2S rm L cyl C rm . The factor Lcyl also is not really defined. Theoretically only the length measured from the point where the feed enters the pond to the liquid overflow is the socalled clarification length that should be taken into account. In most cases the length is taken as the cylindrical length which is a good estimation because the feed enters almost at the joint of cylindrical and conical part. In case Sigma is only based on the cylindrical part of the Decanter using the clarification (or cylindrical) length now the job is done. Some Decanter producers add a certain part of the cone to the system. In this case we are talking about the submerged cone section. The length of this section depends on the bowl diameter, the solids discharge diameter, the half-cone angle and, of course, the weir setting.
There is still a discussion which is the correct Sigma equation with respect to the individual application. The question is. Anyhow the submerged cone only contributes to Sigma if the flow path for the liquid between the flights is not restricted by a baffle or equivalent devices. Because if a restriction (blockage) exists, the submerged part of the cone is hydraulically separated from the cylindrical part, e.g. see Fig.3 above. Now let us look to the results of calculating Sigma using the different equations. First of all, we have to choose a model Decanter: Clarification length, mean: 1400.0 mm Inner bowl diameter: 500.0 mm Conical length, total: 500.0 mm Solids outlet diameter 300.0 mm Conical length, submerged: 480.0 mm Set weir diameter: 420.0 mm Total bowl length: Cylindrical length Clarification length, min: Clarification length, max:
2200.0 mm 1700.0 mm 1300.0 mm 1500.0 mm
Half cone angle: Bowl speed:
10° 3,000 rpm
Based on the data above, the settling area and the equivalent settling area (Sigma) have been calculated. See the results [7]: Settling area, Sokolov: -----dto.-----, Travinski: -----dto.-----, Producer A -----dto.-----, Producer B -----dto.-----, Producer C -----dto.-----, Producer D -----dto.-----, Producer E
2.67 m! 2.69 m! 2.45 m! 3.38 m! 1.72 m! 2.19 m! 2.11 m!
rel.: 1 rel.:1,01 rel.:0,92 rel.:1,27 rel.:0,64 rel.:0,82 rel.:0,79
Sigma, Sokolov: -dto.-, Travinski: -dto.-, Producer A -dto.-, Producer B -dto.-, Producer C -dto.-, Producer D -dto.-, Producer E
6,716 m! 6,219 m! 5,184 m! 8,317 m! 3,624 m! 4,133 m! 5,098 m!
rel.: 1 rel.:0.93 rel.:0.77 rel.:1.24 rel.:0.54 rel.:0.62 rel.:0.76
From the results of this basic calculation one can obtain that, depending on the equations used, Sigma varies with factor 2.3. This of course makes it difficult to compare Decanter data given by different producers. A good way to escape is to agree on a certain set of assumptions and equations on which the up-scale process is based. But calculating for the capacity based on the Sigma-Theory and Stokes’ law is not applicable to all products / applications. The boundary condition of single particle settling as it is needed for stokes can’t be assumed as given for a lot of applications. Scale-up via pond volume / G-Volume I many cases its the pond volume, better the residence time of the liquid in the clarification zone that is the basis for scale-up. This is especially true for applications involving fines or soft “solids” that are handled with deep-pond Decanters. The pond volume /residence time is easy to calculate. It only needs the agreement on the clarification length. Multiplying the pond volume by the acting G-Force (as above depending on an agreement on the relevant radius) leads to the criteria of G-Volume. Basically this is a combination of residence time and driving force. A word on the liquid velocity In case of a fully helical conveyor the cross sectional area for the flow is the pitch multiplied by a fraction of the pond depth. There are a number of approaches in literature to calculate for the resulting flow pattern. In most cases the flow will show a turbulent flow pattern. The velocity of course acts on the sediment by applying a drag force on the settling particles. This may cause a disturbed settling or (worst case) a dragging out of fines by the liquid. This problem could be overcome by using a ´´quasi axial flow conveyor`` that enables for a lower liquid velocity and in consequence less drag force (see next chapter). Cone geometry Before discussing about the cone, let’s look to the different de-humidifying regimes and transport properties of the solids / sediments [7]. Coarse, hard-shaped particles tend to de-humidify via drainage through the cake porosity.
The drained liquid flows into the pond. This type of solids usually do not show any transport problems while conveyed up the cone to the solids discharge. Here basically both are possible, a flat cone with long cake residence on the beach or a steep cone. It depends on the de-humidifying kinetics that can be calculated based on information on the capillary structure of the system (Bond-diagram). Conveyor flight
Product: - crystalline particles - rel. low fitration resistance
Decanter bowl: - shallow pond
Dewatering: - by drainage - liquid flows back into the pond through the cake porosity
Figure 5: De-humidifying by drainage through the cake
Pasty sediments with high filtering resistance, let’s call them sludges, are de-humidified in a totally different way. The key to success here is sediment compression. This means the forced decreasing of sediment porosity by a force given by the sediment mass in combination with the relevant G-Force. This is normally done using a deep-pond Decanter allowing for a relatively high sediment layer. The liquid getting free by sediment compression flows to the sediment surface. This de-humidifying regime in most cases goes together with weak conveying properties of the sediment. Here a flat half cone angle is necessary if there is no hydraulic help via a baffle (see next chapter). Conveyor flight
Decanter bowl: - deep pond
Product: - pasty products - high filtration resistance
Dewatering: - by sediment compression - liquid flows to cake surface and (if possible) meanders back ito pond.
Figure 6: De-humidifying by sediment compression
As the total bowl length is given, a steep cone is of advantage relating to clarification length. But it requires either good conveying properties of the sediment or a “helping hand” by hydraulics (baffle disc). 3. THE CONVEYOR
The conveyor is the heart of the Decanter Centrifuge. It has to facilitate the solids / sediment transport along the cylindrical length and up the cone, to allow for a nondisturbed settling and good / fast de-humidifying [9 – 13]. Not an easy task! Multi flight conveyor Again let’s start with coarse solids exhibiting low filtration resistance. In this case the dehumidifying on the cone could be boosted by decreasing the height of cake “triangle” in front of the flights. This can be achieved by using double- or even triple flight conveyor.
Assuming the solids load and the differential speed being constant, a double flight conveyor will cut the cake height in front of the flight by half. Its easy to imagine that this leads to faster de-humidifying.
Single flight
Double flight
Figure 7: Single- and double flight conveyor.
Employing a cake baffle We have already discussed the transport properties of different sediments and stated that some sediments need a helping hand to smoothly move up the cone. This “hand” could be given by a cake baffle and subsequent negative weir setting. Negative weir setting means setting the weir on a smaller diameter than the solids discharge. In this case the subtraction (weir diameter - solids outlet diameter) results in a negative figure. In consequence the incoming feed could take the short way out via the cone and the solids discharge. So no separation will occur. But if there is a baffle blocking the solids transport at the cylinder cone fuselage the story is different. The settling solids in the cylindrical part will be transported up to the baffle by the conveyor, accumulating in front of the baffle disc. This leads to a “sealing” of the annular gap between the baffle and the bowl wall. Now there is no chance for the feed to take the short way out. But how is the sediment transported across the baffle? This transport is facilitated by means of a hydraulic pressure difference on both sides of the baffle. Caused by the negative weir setting the pressure on the cylindrical side is higher than on the conical side. This effect will press the sediment through the annular gap and up the cone. Of course a pre-requisition for the function of this system is a soft, pasty sludge that de-humidifies by compression. Consequences using a baffle Using a baffle leads to a number of consequences: x No beach and no de-humidifying on the cone. x Free liquid on the cone side will be discharged with the sediment; no other escape. x Differential speed must be set to enable for the “sealing”. x Low 'n; only limited by the scrolling capacity on the cylindrical part. x High sediment build-up in front of the baffle Æ sediment compression x Potential wash out if 'n does not fit. x At low 'n the solids can move faster on the cone than the conveyor. x Reduced torque level compared to an operation without baffle.
Figure 8: Scroll with cake baffle disc
Quasi axial flow conveyor Some applications require a low speed of the liquid moving along the clarification zone. This could be facilitated by employing a so-called quasi axial flow conveyor. The axial flow is achieved by means of windows, cut into the conveyor flights. The liquid now is able to take the direct way from the inlet to the weir overflow. The free flow section is given by the sum of the window areas over 360°. Depending on the window size that in most cases increases towards the front end, this section is much bigger than the free flow section between the flights (especially when taking into consideration the theory of moving and stagnant liquid layer). Using a quasi axial flow conveyor possibly contributes to the separation of soft easy to break (by shear stress) and easy to drag-out particles, e.g. sludge flocks.
Figure 9: Quasi axial flow conveyor
The surface question Basically the surface of the flights and the feed zone must comply with the mechanical properties of the feed suspension and the solids / sediment respectively. In addition some applications demand surfaces that cope with hygienic demands. In most cases the feed zone and the flights from behind the feed zone up to the solids discharge are subject to extensive wear. Therefore a wear protection is often needed, e.g. flame-spray of Tungsten Carbide, Tungsten Carbide Tiles or Ceramic Tiles. The appropriate hard surfacing must be chosen combining the mechanical (erosion) and chemical (corrosion) demands.
Figure 10: Quasi axial flow conveyor equipped with TC-Tiles
No matter which surface is chosen, the total friction of the cake / sediment on the flights must be less than that on the bowl-wall (on the primary layer). If this rule is not obeyed, the consequence will be massive transport problems of the cake / sediment and its accumulation inside the machine. The rake The rake (the angle of the conveyor flight with the cone surface) is of significant influence with respect to the transport forces for the cake / sediment. A positive rake tends to lift the cake from the bowl (not conveying “towards the cone wall”) thus reducing the torque by decreasing the friction. A negative rake will ”produce” by increased friction. It will act on compressible sludges by increasing the compression force. It depends on the friction balance between the cone surface and the flight surface whether this is positive (better dehumidifying) or negative (transport problems) …
Positive Rake
Negative Rake
Figure 11: The rake
4. CENRATE OUTLET At the front end, the clean liquid leaves the Decanter via a “discharge device”. Of course, the separation job is done as the liquid reaches the outlet. But the design of the outlet contributes to the variability of the machine, to the handling of the centrate e.g. if a contact with the atmosphere is a problem and to possible energy saving [10, 13, 14]. No doubt, for many applications the traditional weir overflow is totally sufficient.
Figure 12: traditional weir overflow
The weir-type outlet leads to an intensive mixing of the centrate with air / atmosphere thus possibly creating a gas/liquid flow in the down-stream pipe or (worst case) foam problems. We have also to keep in mind that the weir overflow exhibits a crest depending on the weir design, the overflow length and of course the volume flow.
An alternative is given by a paring device, e.g. a paring tube or paring disc (centripetal pump). The centrate discharge via a paring device gives three main advantages: Gas-free discharge, inprocess adjustable pond depth and discharge under pressure. The possibility of setting the pond depth is of interest especially for interface positioning in case of 3-phase separation. Disadvantages are given by the fact that the paring device, respectively the needed paring chamber, could be clogged with sediment in case of insufficient settling and that there is a need of energy for this system that has to be fed via the main drive.
Heavy phase
Light phase
3-phase, paring device Figure 13: Paring disc, 3-phase application
The latest development with respect to liquid discharge is the attempt to recover energy by “directed” outlet. This means, that the liquid leaving the decanter is directed to a tangential outflow thus resulting in a repulsion that avoids the traditional energy loss on the outlet. The tangential outlet could be facilitated by means of 90° weir-cups or e.g. an outlet via tangential nozzles. 5. CONCLUSIONS A Decanter is not just a Decanter. It could be a highly specialized separation device tailored to give the best performance in an individual application. All major parts / functions can be specified to comply with the demands of the products to be handled. It is the challenge of the co-operation between user and producer to specify and design a Decanter giving maximum performance in a specific application but not being limited to this one and only separation task. As one can see, meeting this challenge needs a high grade of expertise on both sides – and an open communication. 6. REFERENCES
[1] [2] [3] [4] [5] [6]
PIERALISI Group, Applications – List, Internal Paper 2005 Hebb, M.H. , Smith, F.H. Centrifugal Separation Encyclopaedia of Chemical Technology, 1st Edition, Vol. 3, 1949 Faust, T., Untersuchnugen über die Strömungen und Absetzvorgänge in der Klärzone von Dekantern. Dissertation Universität Stuttgart, 1983 Langeloh, T., Einfluss des Schleppkrafteffektes auf die Klärung in Dekantierzentrifugen, Dissertataion Un iversität Karlsruhe, 1986 Sokolov, W.J., Moderne Industriezentrifugen, VEB-Verlag Technik 1971 Trawinski, H., Kapazität, Trenneffekt und Dimensionierung von Vollmantelschleudern, CIT, 31 (1959)
[7] [8] [9] [10] [11] [12] [13] [14]
VDI Seminar Industriezentrifugen, Tagungshandbuch, 2005 Zum Einfluss der Zulaufgeometrie und der Viskosität auf dir Strömung min Überlaufzentrifugen, Dissertation Universität Dortmund 1977 Madson, N.F. Decanter Centrifuge, US Patent 4828541, May 9, 1989 La Montagne, P. Centrifuge employing variable height discharge weir, US Patent 4575379, March 11, 1986 Lee, C.Y., Centifuge Apparatus, US Patent 3795361, March 5, 1974 Locke, D., Trueman, J.W., Screw conveyor for centrifuges, GB Patent 2273253, June 15 1994 Records, A., Sutherland, K., Decanter Centrifuge Handbook, 1st Edition 2001, Elsevier Science Ltd. Beyer, H.-J., Handbuch der Mechanischen Fest-/ Flüssigtrennung, Vulkan-Verlag 2004
