DYNAMIC MODEL OF A DISC FILTER Jouni Savolainen, Sami Tuuri and Kaj Juslin VTT Automation, P.O. Box 1301, FIN-02044 VTT Finland Jouni.Savolainen@vtt.fi 1. Abstract The purpose of this paper is to describe the design and implementation of a dynamic simulation model of a rotary disc filter to the APMS simulation environment. The construction and operation of a disc filter is described. The APMS simulation platform is introduced. The calculation principles and methods used in the model are described. The test runs and their results as well as the conclusions are presented. The model was found to respond in a qualitatively correct manner to feed flow and consistency disturbances. 2. Introduction The present work deals with the modeling and simulation of a rotary disc filter. The model is based on empirical operating curves, the so called leaf-test curves, and mechanistical balance equations. The model was built on the APMS simulation software developed by The Technical Research Centre of Finland (VTT). The aim of the work was to model the disc filter so that the relevant dynamic properties can be simulated. This work concentrated in modeling such disc filters that are used in the pulp and papermaking industry. /1/ 3. The Disc Filter A disc filter is a continuously operating rotary filter. The basic construction of a disc filter includes the filtration discs, a horizontal vat, a center shaft and piping. The filter discs are attached to the hollow center shaft and they rotate partly submerged in the vat. Each disc is divided into several sectors. At one end of the center shaft there is motor which turns the shaft and consequently the disc at a rate of usually less than 2 r/min. At the other end of the filter the center shaft is connect to the filtrate output piping, which usually are drop legs. The construction of a typical rotary disc filter according to Wakeman et al. /4/ is depicted in Figure 1. Figure 1 Typical construction of disc filter /3/ During the operation of the filter each sector in a disc is submerged in the liquid part of the time. During this time the actual cake formation and filtration is carried out. The cake is formed on the outside surface of the hollow discs. Usually there are two or three filtrate flows depending on the construction of the device. These are named cloudy, clear and super clear corresponding to their solids content. The division of the filtrates is accomplished by a special valve fitting at the end of the center shaft. With this valve it is also possible to alter the relative amounts of the filtrates. The driving force for the filtration can come from different sources. The first and most commonly used driving force is vacuum applied to the inside of the discs. This vacuum can be attained for example by drop legs. The second possible driving force used is overpressure on the outside of the discs ie. in the vat, which then is sealed from the atmospheric pressure. When a sector of a rotating disc rises above the liquid level the vacuum is turned off by the filtrate valve and the filter cake is removed either by water showers or by scrapers. After this the disc sector may washed and then it starts the cycle again. In the papermaking industry disc filters are used mainly for two purposes. The first use is as a saveall filter in the fiber recovery system of a paper machine. There its purpose is to recover valuable fibers and other solids from the white water coming from the paper machine's wire pit. The recovered fibers are then returned into the mixing chest in the approach system of the machine as raw material for paper. The filtrates are lead to different places according to their solids content. The cloudy filtrate is mainly recycled straight back to the feed of the disc filter. Part of the clear filtrate is used as shower water in the disc filter itself and part of it is pumped to the showers of the paper machine and elsewhere in the mill. The second use of a disc filter is as a thickener in the broke handling of the paper machine. The purpose of the broke handling system is to treat the broke so that it can be reused in the paper making process without runnability or paper quality problems at the paper machine. The broke from the pulpers is thickened in the broke thickener (disc or drum filter) from about 2.5-3.5 % to 4.0-5.0 % before being pumped to the mixing chest. 4. APMS Simulation Software The APMS (Advanced Pulp and Paper Mill Simulator) is an extension to the APROS (Advanced PROcess Simulator). APROS is a dynamic simulation software developed jointly by VTT and Fortum Engineering Ltd /2/. With APROS it is possible to simulate a variety full-scale industrial processes and their automation. Using different unit model packages APROS can used to simulate nuclear power plants, combustion power plants, pulp mills and paper mills. The last two constitute APMS. Currently APMS includes models for most of the paper mill unit operations /3/. The pulp mill models are under development. The models are configured with a graphical user interface by placing units on the screen, connecting them and filling in properties with dialogue windows. With APMS it possible to analyse both transient and steady state situations. The models can be saved into a snapshot file which contains all the information concerning the model and its current state. The snapshot files can be read back to simulation software in case the user wishes to revert back to a previous situation. 5. Calculation principles and implementation of the disc filter model The disc filter was modeled using empirical operational curves, so called leaf-test curves. These curves describe the drainage capacity of the filter and the quality of a filtrate as a function of the rotation speed and the consistency (percent of dry solids in the suspension) in the filter vat. An example set of operational curves is shown in Figure 2. Figure 2 Operational curves of a disc filter In the model the operational curves were approximated by a straight line through the origin and a nominal operating point which is calculated from the values specified by the user at the time of adding a new disc filter model into the simulation experiment. The values the user must specify are the nominal flows and concentrations of the filtrates as well as the feed. Also the nominal rotation speed must be defined. During the simulation the filtrate mass flow is calculated using the linearized operational curves plus a term for the pressure difference over the filter cake. The calculation of the concentrations of the filtrate flows uses only the operational curves. The recovered fiber mass flow and concentrations are then calculated from mass balances. 6. Test runs and results 6.1 The test model The test model which is depicted in figure 3 included the disc filter, three filtrate tanks, a recovered fiber tank and the pipe lines including pumps and control valves. The process component connections in the model were made to resemble those of the fiber recovery system of a paper machine. There are three streams in the feed line which are mixed before the filter. The main flow comes from the wire pit of the paper machine and is called white water. Part of the recovered fiber is mixed, as the so called sweetener stock, with the white water. In real life this is done in order to enhance the filtration properties of the stock in the disc filter. The sweetener need not necessarily be the recovered fiber but other stocks will also do. The third stream in the feed side of the filter is the recycle stream from the cloudy filrate tank. There is also a third recycle stream in the process which recycles part of the clear filtrate to the mat removal and disc washing showers of the filter. The length of the filter vat was 8.0 m and it's diameter 4.0 m giving a total volume of 100m3 where as the filtrate tanks and the recovered fiber tank had each a diameter of 1.13 m and height of 5.0 m giving a volume of 5.0 m3. In the disc filter the filtrates where divided in the ratio: 40% cloudy, 45% clear and 15% super clear. In addition to the feed, filtrate and fiber recovery piping an overflow pipe was connected to the filter. The height of the overflow from the filter bottom was set to 3.5 m. There were overflows from all the other tanks as well, all of them set to 4.5 m. In the model there were also two control loops. The first control loop was used to control the liquid level of the disc filter by manipulating its rotation speed. The liquid level setpoint was 3.0 m. The second control loop was used to control the flow of the sweetener stock which is mixed with the white water. The sweetener flow setpoint was 10.0 kgs-1. The model was tested by making disturbances to the white water feed flow and consistency and by recording the responses in the filtrate mass flows and consistencies, in the liquid level of the vat and in the rotation speed of the filter. In the test experiment the flowing substances were water and one fiber component although APMS does support multiple components in the flow network. Figure 3 The test model 6.2 Feed flow increase In the first test run the feed mass flow was abruptly increased from 178 kgs -1 to 326 kgs1 . This was accomplished by increasing the feed pressure from 420 kPa to 500 kPa. The behaviour of the liquid level after the feed mass flow is depicted in the primary axis of figure 4. In the figure the rotation speed is shown on the secondary axis. Figure 4 Effect of feed mass flow increase on the liquid level First the liquid level starts to increase but as its value deviates from the setpoint of 3.0 m the level controller starts to increase the rotation speed of the filter. In accordance to the operational curves of figure 2 the filtrate mass flows increase in value as shown in figure 5. This in turn returns the liquid level back to the setpoint value. Figure 5 Filtrate mass flows The effect of the feed mass flow on the consistencies in the disc filter vat and in the cloudy filtrate flow are shown in figure 6. Figure 6 Consistencies of the vat and cloudy filtrate The behaviour of the consistency trends can be explained using the operational curves of the filter. Because the level controller starts to increase the filter's rotation speed the quality of the filtrates decreases ie. their consistencies increase. This is due to the fact that the cake forming and filtration time decrease and the filtrate cake does not retain the solids as effectively as with a lower rotation speed. The increase in the vat consistency is due to the recycle stream from the recovered fiber tank to the feed of the filter. The consistency of the vat does not exhibit such an abrupt increase as the cloudy filtrate consistency because of the additional mixing in the fiber tank. 6.3 Feed consistency increase In the second test run the white water consistency was increased in step-like manner from 0.4 % to 0.8 %. This caused a gradual increase in the consistency of the filter vat which then affected both the filtrate flows and their consistencies. Also the consistency of the recovered fiber increased. The effect of the disturbance on the vat and fiber consistencies is shown in figure 7. The feed and vat consistencies are on the left vertical axis and the recovered fiber consistency on the right vertical axis. Figure 7 Feed and vat consistencies First the feed consistency increases rapidly and then smooths down. After a while it starts to increase gradually due to the increase in the consistency of the recycled fiber. Also the filtrate consistencies increase which add to the feed consistency via recycle streams. The mass flows of the filtrates decrease when the consistency of the feed flow increases as indicated by the operational curves of figure 2. This decrease in the flows causes the liquid level in the vat to rise. The level is then returned back to its setpoint by the level controller by increasing the rotation speed. The rotation speed evens out when the setpoint is again reached causing the filtrate mass flows to stabilize. These effects are shown in figures 8 and 9. Figure 8 Filtrate mass flows Figure 9 Liquid level and rotation speed 7. Conclusions The unit model of a disc filter developed here was integrated as a part of the APROS/APMS system and tested there. In the tests two step changes were made to the input of the process and the dynamic behaviour of the system was monitored. From the tests conducted it can be deduced that the system gives qualitatively correct responses in all simulated cases. To get also quantitatively correct responses the model is possible to tune to equipment specific industrial measurement. 8. References 1. Savolainen, J., Dynamic model of a disc filter, Master's Thesis, Department of Chemical Technology, Helsinki University of Technology, 1999, Espoo, 76+24 p. (In Finnish) 2. Silvennoinen, E., Juslin, K., Hänninen, M., Tiihonen, O., Kurki, J., Porkholm, K., The APROS software for process simulation and model development, VTT Research Reports 618, Technical Research Centre of Finland, Espoo 1989, 106 p. 3. Niemenmaa, A., Lappalainen, J., Laukkanen, I., Juslin, K., An Advanced Platform for Dynamic Simulation of Paper and Board Mills, SPCI '96. 4. Wakeman, R.J., Tarleton, E.S., Filtration Equipment Selection Modelling and Process Simulation, Elsevier Advanced Technology, Oxford 1999, 446 p.