EP3866954A1

EP3866954A1 - Membrane filter system and method for controlling same using fuzzy logic and/or artificial neural networks

Info

Publication number: EP3866954A1
Application number: EP19789927.1A
Authority: EP
Inventors: Andreas HIPPEL; Joerg Zacharias; Daniel NITSCHE; Maximilian FRAMMELSBERGER
Original assignee: Krones AG
Current assignee: Krones AG
Priority date: 2018-10-19
Filing date: 2019-10-15
Publication date: 2021-08-25
Also published as: DE102018218440A1; CN113260447A

Abstract

The present invention relates to a device for the filtration of fluids, in particular beer, having: a first filtration unit, which can be controlled independently in an open- and/or closed-loop manner and has at least one membrane filter module; a second filtration unit, which can be controlled independently in an open- and/or closed-loop manner and has at least one membrane filter module; and at least one control unit, wherein the control unit is designed to adaptively control a loading of the second filtration unit on the basis of at least one process parameter of a filtration with the first filtration unit.

Description

Membrane filter system and method for regulating the same

Field of the Invention

The present invention relates to a membrane filter system for filtering beer and a method for controlling and / or regulating such a membrane filter system.

State of the art

In the production of beer, the beer is filtered after fermentation / maturation, in particular to remove yeast from the beer. The membrane filtration of beer has been an increasingly used technology for some years. The crossflow method is used in particular, in which the unfiltered beer, i.e. the unfiltrate is passed in a circuit through the membrane filter and the filtrate is withdrawn from the membrane filter. For example, plastic hollow fibers or ceramic filter cartridges with microfiltration pores are used as the membrane. The membranes used are therefore also referred to as microfiltration membranes.

During the filtering process, substances to be filtered settle on or in the membrane, which increases the filter resistance and ultimately reduces the efficiency of the filtering process. At the same time, the concentration of the retained components in the unfiltrate, in particular the yeast concentration, as well as the concentration of proteins, bitter substances and polysaccharides, etc., increases over time. In the course of the filtration, the unfiltrate concentrates so strongly that the filtering process must finally be stopped in order to dispose of the concentrate, whereby the membrane can be cleaned. Also through membrane fouling, i.e. the accumulation of soiling, for example of yeast, proteins, hemicellulose or the like, on a membrane, the membrane becomes increasingly clogged, so that cleaning, for example by backwashing with alkali and / or acid, is required.

To control the filtration, the transmembrane pressure can be observed, for example, and the filtration can then be stopped at predetermined limit values, the concentrate being pushed out and / or the concentrated unfiltrate being diluted with fresh unfiltrate. The blocking of the membrane can be delayed by backwashing with filtrate, in particular with beer, and / or water. However, the use of predetermined limit values means that the reaction only takes place when the membrane is already blocked, so that blocking of the membrane cannot be prevented at an early stage. The membrane filters for beer filtration in the prior art therefore only have a short service life.

In addition, the membrane filtration of beer, especially the service life of the membrane filter before necessary cleaning, depends heavily on the raw materials of the beer Recipe and fermentation. The duration of a filtration cycle, ie the service life of a membrane filter, before cleaning, for example by backwashing with certain backwashing media or by cleaning-in-place (CIP), is very dependent on the beer to be filtered.

The result is a not insignificant planning uncertainty for the operator of the filter system, for example the brewery, whereby undesired early, long or inadequate cleaning processes of the membrane filters are often carried out. This usually leads to longer production times and increased operating costs.

Due to the large number of process parameters that affect membrane fouling and blocking of the membranes and are difficult to capture in a model, the usual approach when operating filter systems with membrane filters is to use fixed program sequences with regard to the filter processes and / or the cleaning processes. For example, fixed limit values for the transmembrane pressure are specified as termination criteria for the filter process. Likewise, overflow speeds and filtration flows of crossflow filtration are frequently specified constantly or increased disproportionately quickly when the filter becomes clogged.

In addition, it has so far only been possible with control systems of the prior art to predict the duration of the filter process linearly in advance or during the filtration, in order to initiate measures in good time by which the filtration can be extended or optimized. However, a linear prediction model generally deviates from the real situation, since the filtration process is linear at best in the first half of a filtration and then usually decays exponentially.

Due to the fundamental dependence on the production schedule of every brewery, this leads to unsatisfactorily small amounts of beer being filtered and the utilization of the brewery not being optimal.

Furthermore, due to the rigid / inflexible process sequences (for example, fixed times or fixed flows or fixed flow directions), the membrane filter cleaning is not carried out optimally. The duration of membrane filter cleaning is generally too long and can only be reduced by using more chemicals (for example, a higher concentration of lye or oxidizing agent). This increased use of chemicals results in higher operating costs and has a negative impact on the life of the membrane modules, which also increases the costs for the operator.

If the membrane filter is not cleaned optimally, the subsequent filter process is often shortened. In total, the elusive and difficult Modelable framework conditions of the beer parameters at the beginning and during the ongoing filtration as well as the rigid control of the filter systems, the automation of an optimal beer filtration. The control becomes more complex the more filtration units are to be used in parallel.

It is therefore the object of the present invention to provide an apparatus and a method for the filtration of beer which overcome the above-mentioned disadvantages of the prior art. In particular, the service life of the membrane filter should be extended, the cleaning of the membrane filter should be optimized and the control of the filter system should be automated. Despite changing beer parameters, optimal filtration results should be achieved, particularly with regard to productivity and production costs. Furthermore, an automated control or regulation of a filter system with several filtration units is to be made possible.

Description of the invention

The above-mentioned objects are achieved by a device for the filtration of fluids, in particular beer, comprising: a first, independently controllable and / or regulatable filtration unit with at least one membrane filter module, a second, independently controllable and / or regulatable filtration unit with at least one membrane filter module, and at least one control unit, the control unit being designed to adaptively control an occupancy of the second filtration unit as a function of at least one process parameter of a filtration with the first filtration unit. Alternatively or in addition, the control unit can be designed to adaptively control an occupancy of the first filtration unit as a function of at least one process parameter of a filtration with the second filtration unit.

In addition to beer, other fluids, for example water, milk or other liquid foods, can also be filtered with the device according to the invention. The filtration can be carried out as dead-end filtration or crossflow filtration. In crossflow filtration, the unfiltered fluid, in particular beer, ie the unfiltrate, is generally passed in a circuit through the membrane filter, the filtrate being drawn off from the membrane filter. The unfiltrate flows along the membrane, the filtrate usually exiting perpendicularly to it. For example, plastic hollow fibers or ceramic filter cartridges with microfiltration pores are used as the membrane. Depending on the pore size, one speaks of microfiltration or ultrafiltration. The pore size for beer filtration is in particular in a range from 0.1 to 1 pm, in particular from 0.4 to 0.6 pm. Polyether sulfone, for example, can be selected as the material for a hollow fiber membrane. In the case of membrane filters, a large number of hollow fiber membranes can be bundled or combined to form a membrane filter module, the hollow fiber membranes being able to be integrated in a stainless steel pressure tube. Several membrane filter modules can be combined in one filtration unit. The individual membrane filter modules can be connected to one another, in particular in a row, or also partially or completely parallel to one another, by means of appropriately designed connecting lines or pipes, control or regulating valves, pumps and other elements of filtration units known per se.

According to the invention, the first and second filtration units can each be controlled and / or regulated independently. For this purpose, the first and the second filtration unit can have the above-mentioned elements, in particular controllable valves and pumps, which allow independent control and / or regulation of the filtration with the respective filtration unit. In a variant, each filtration unit can have its own control and / or regulating unit for this purpose, which can be designed, for example, in the form of a programmable logic controller. Other elements known per se, such as a buffer tank for the unfiltrate, a buffer tank for the filtrate and one or more integrated CIP modules and / or backwashing lines, can be provided as part of the first and / or second filtration unit. When using the crossflow method, the first and the second filtration unit can in particular have separate circuits for the unfiltrate.

The first and the second filtration unit can be connected, for example via a controllable three-way valve or a valve node, to a feed line for the unfiltrate, in particular to a main buffer tank for the unfiltrate.

It goes without saying that equivalent filtration units can be provided which can be designed and controlled in accordance with the further developments described below.

According to the invention, at least one control unit is also provided which is designed to adaptively control an occupancy of the second filtration unit as a function of at least one process parameter of a filtration with the first filtration unit. Additionally or alternatively, the control unit can be designed to adaptively control an occupancy of the first filtration unit as a function of at least one process parameter of a filtration with the second filtration unit. In very general terms, the control unit can be designed to adaptively control the occupancy of a specific filtration unit from a plurality of filtration units as a function of at least one process parameter of a filtration with at least one other filtration unit. A multiplicity of independently controllable and / or regulatable filtration units can thus be provided, the occupancy of at least one of the filtration units being adaptively regulated as a function of at least one process parameter of a filtration with at least one further filtration unit. Process parameters of two or more further filtration units can thus also be taken into account when regulating the occupancy of the filtration unit. As described below, the regulation can be carried out by a separate control and / or regulating unit of the filtration unit, the occupancy of which is regulated, by separate control and / or regulating units of the at least one further filtration unit or by a higher-level regulating unit. Furthermore, the occupancy of all filtration units can be regulated adaptively, with process parameters from at least one further filtration unit being taken into account in each case. Such control can be carried out in particular by the higher-level control unit. Finally, process parameters of all other filtration units can be taken into account when regulating the occupancy of all filtration units. In this way, the overall occupancy of the filter system can be optimized.

Here and below, the term filtration is generally to be understood to mean a sequence or sequence of filter processes and cleaning processes of the membrane filter modules involved, it also being possible to provide rest periods or breaks between cleaning processes and filter processes in which the filtration unit is in standby mode located. In contrast to the term filtration, the actual process of filtering the unfiltrate is referred to here and below as the filter process. In the cleaning process, at least one membrane filter module, in particular all membrane filter modules, of the respective filtration unit are cleaned. A variety of different cleaning methods can be used, which are described in more detail below.

The occupancy of a filtration unit is to be understood here and below as the time sequence of the filter processes and cleaning processes mentioned above, potentially interrupted by standby phases. For example, the service life of a filtration unit results from the ratio of the time occupied by filter processes to the total filtration time. In addition to the pure filter processes and cleaning processes, other customary process phases, for example the pretensioning of the membrane filter modules and the retraction of the filtration unit, are also taken into account in the adaptive regulation of the occupation of the filtration unit described below. In contrast to, for example, the pretensioning or the shutdown, the respective process duration of the filter processes and cleaning processes is not defined in advance, but is, as described in more detail below, adaptively adapted, that is to say regulated, depending on one or more process parameters of a filtration with one or more filtration units. In the prestressing step, the membrane filter system, usually with carbon dioxide, can be set to operating pressure and prepared for the filtering process step. Leaving is generally the step after the actual filtering. The membrane module can be emptied / freed from the concentrated unfiltrate and the filtrate can be pushed out, the appropriate steps for cleaning being initiated.

The term process parameter is to be understood in general here and below and can include any type of parameter that belongs to the filtration with the respective filtration unit. For example, are hereby offline, i.e. separately from the operation of the filtration unit, for example in the laboratory, includes parameters measured, in particular with regard to the fluid to be filtered. In the case of beer, for example, the type (light, dark, Pils, Bock, wheat beer, etc.), the season, the season, the region from which the raw materials come, the malt type (for example with regard to protein content or hemicellulose), the raw fruit (for example corn, rice, barley, sorghum), enzymes used, and the like to process parameters that influence the filtration, in particular the fouling of the membrane filter. In the broadest sense, this also includes events such as events that determine the market need for beer or the dependence on a particular operator of the filter.

The technological process parameters, on the other hand, include directly measurable values, such as the yeast cell count, the turbidity, high-gravity brew, the fermentation time, the storage time, the use of top or bottom-fermenting yeast, the viscosity, the density, the general filterability, the pH and the like. The technological process parameters also include parameters that were previously measured offline, ie in the laboratory, for example the content of protein, completely soluble nitrogen, MgS0 ₄ - precipitable nitrogen, free amino nitrogen, polyphenols, anthocyanogens, glucans (a, ß, gel) , Original wort, extract, the degree of fermentation, the degree of final fermentation, the color, iodine values, in particular photometric iodine values, of the unfiltrate, the number of foams, bitter units, the alcohol content or the like.

Here and below, the term process parameters also includes parameters that are directly influenced by the control or regulation of the respective filtration unit. In particular, this also includes control parameters such as the flow rate of the unfiltrate in the circuit, in particular the overflow speeds of the filter membranes, the volume flow of filtrate, the pressure level on the unfiltrate side or on the filtrate side, the transmembrane pressure, the degree of concentration, the cooling temperature, the speed and volume flow of a recirculation, flushing volumes, as well as concentration and type of CIP media used. Other process parameters are the parameters: temperature of the fluid in a filter inlet, pressure of the fluid in the filter inlet, pressure of the fluid in a filter outlet, pressure of a filtrate, differential pressure of the fluid between the filter inlet and the filter outlet, differential pressure between the fluid in the filter inlet and the filtrate, Volume flow of the fluid supplied in the filter inlet, volume flow of the filtrate, flow rate of the fluid supplied in the filter inlet, flow rate of the filtrate, yield of the filter, operating time of the filter, filter life, filter runtime, turbidity of the fluid in the filter inlet, turbidity of the filtrate, concentration gradient Particles to be separated in the filter inlet, thickness of a cover layer on the filter, density of the cover layer on the filter, adsorption of particles in the filter body itself, filter resistance of the filter, filter throughput, filter exclusion limit, degree of hardness of the fluid in the filter inlet, degree of hardness of the filtrate, electrical conductivity of the fluid in the filter inlet, electrical conductivity of the filtrate, concentration of a salt in the fluid in the filter inlet, concentration of the salt in the filtrate, concentration of an ion in the filter inlet that is decisive for membrane fouling, concentration of the ion decisive for membrane fouling in Filtrate, number of a filtration cycle, backwashing resistance of the filter, volume flow of a backwashing fluid, flow rate of the backwashing fluid in a backwashing inlet, turbidity of the backwashing fluid in a backwashing outlet, differential pressure of the backwashing fluid between the backwashing inlet and the backwashing outlet, duration of a backwashing process and the lifetime of the filter, as well as their life time Deviations from predetermined reference curves.

The term process parameters here in particular also includes parameters output by a control and / or regulating unit of a filtration unit or the regulating unit of the device, in particular with regard to actions which are used to intervene in the filtration. As a non-exhaustive list, for example, the time, strength and duration of backwashing steps, e.g. with beer and / or water, intermediate cleaning, e.g. short cleaning of the membrane with lye and additives, as well as main cleaning, e.g. intensive cleaning of the membrane and the periphery such as the concentration tank, number and type of cleaning steps, type and amount of chemicals used for cleaning (acids, alkalis, additives, oxidizing agents, enzymes, and the like), running times of filter processes and cleaning processes, termination of a filter process Shortening a filtering process, lengthening or shortening a CIP cleaning process, starting or preparing a filtering process, standby, time and duration of a sleep state, an optimized occupancy plan, changes or adjustments in the brewing process and the like.

The control unit can be designed as a programmable logic controller and, in particular, in addition to a process unit and a memory unit, one or more sensors for determining at least one process parameter of a filtration with the first Have filtration unit. The control unit can be provided as a higher-level control unit of the device, which receives data from separate control and / or control units of the filtration units, processes them and transfers processed data to one or more of the separate control and / or control units. Instead of a higher-level control unit, however, the separate control and / or regulating unit of the second filtration unit can also be used for the aforementioned adaptive control of the occupancy of the second filtration unit. In this case, the at least one process parameter of the filtration with the first filtration unit can be transmitted from a corresponding control and / or regulating unit of the first filtration unit via data lines or wirelessly to the control and / or regulating unit of the second filtration unit.

Because at least one process parameter of the filtration with the first filtration unit is taken into account when regulating the occupancy of the second filtration unit, the control or regulation of the filtration with the first and second filtration units is no longer completely independent of one another. Rather, the at least one process parameter is taken into account in the control of the filtration with the second filtration unit. More specifically, the control unit adjusts a sequence or sequence of filter processes and cleaning processes that are carried out with the second filtration unit, depending on the at least one process parameter of the filtration with the first filtration unit.

In this way, it is possible to react flexibly to the course of the filtration with the first filtration unit, and in particular to the otherwise usual, rigidly predetermined course of the filtration, i.e. a fixed assignment, to deviate if necessary. So far, filtration units have been used to filter according to a specified flow chart and then cleaned. After cleaning, for example a CIP, the corresponding filtration unit is available again for a subsequent filter process. So far, this has been carried out independently of the status of further filtration units, so that deviations from the expected filtration process, particularly with a small number of installed filtration units, could result in bottlenecks in the brewery's output.

By adaptively controlling the occupancy of the second filtration unit as a function of the at least one process parameter of the filtration with the first filtration unit, it is possible in particular to react flexibly to deviations from an expected process duration of a filter process or a cleaning process of the first filtration unit. The control or regulation of the second filtration unit can be adapted such that the occupancy of the entire filter system is optimized. In particular, excessively long standby phases or rest periods of individual filtration units in which these are unproductive can be avoided. In addition, it can always be achieved that at least a filtration unit carries out a filter process so that filtrate is always removed. In this way, for example, an otherwise conventional buffer tank for the filtrate can be made smaller or even possibly omitted.

The process parameter of the filtration with the first filtration unit can include a process duration, in particular residual process duration, a filter process and / or a cleaning process of the first filtration unit, the process duration being predicted by means of the control unit or a separate control and / or regulation unit of the first filtration unit. For this purpose, the control unit or the separate control and / or regulating unit of the first filtration unit can determine one or more process parameters of the filtration with the first filtration unit, in particular measure using one or more sensors. Process parameters that influence the duration of a filter process or a cleaning process or are related to these are preferably determined or measured. For example, an overflow velocity or a volume flow of the unfiltrate, a pressure level of the unfiltrate, a pressure level of the filtrate, the transmembrane pressure or a gradient of the transmembrane pressure as control parameters of the control or regulation of a filter process with the first filtration unit influence the process duration, in particular Remaining process time, the filter process. For example, one end of the filtering process can be delayed by increasing the transmembrane pressure. Furthermore, the above-mentioned parameters of the fluid to be filtered can be taken into account when forecasting the process duration or residual process duration of the filter process. In the same way, parameters can be read from other linked company databases or a cloud from other systems. The above-mentioned parameters of the fluid to be filtered can also be measured offline beforehand and made available to the control unit or measured during the filtering process, for example by sensors arranged in situ.

Findings relating to the relationship between the process duration and the process parameters mentioned above can be incorporated into the forecast of the process duration or remaining process duration of a filter process. In the simplest case, the forecast can be made on the basis of a linear extrapolation of a measured or simulated course of the filter process. As already mentioned above, however, the concentration of the unfiltrate or the fouling of the membrane filter is generally not linear and in particular cannot be represented by simple modeling. There is therefore always a discrepancy between the predicted process duration and the actual remaining runtime of the filter process before at least an intermediate cleaning, for example an intermediate CIP cleaning, has to be carried out. Nevertheless, taking into account the predicted process duration of a filter process of the first filtration unit leads to an optimization of the occupancy of the second filtration unit. For example, a cleaning process of the second filtration unit can be intensified and / or shortened depending on the process parameter of the filtration with the first filtration unit, in particular a predicted remaining process time of a filter process of the first filtration unit. To intensify the cleaning process, for example, a volume flow of a backwashing fluid, a flow rate of the backwashing fluid in a backwashing inlet or a differential pressure of the backwashing fluid between the backwashing inlet and in the backwashing outlet can be increased for a cleaning process by means of backwashing.

Backwashing means that a backwashing fluid is against the direction of filtration, i.e. is led from the filtrate side through the membrane to the unfiltrate side. The filtrate itself, water and / or air and / or a production gas (for example carbon dioxide or an inert gas) and / or cleaning agents such as lye can be used as the backwashing fluid. If only the filtrate is backwashed, the filter process does not have to be interrupted. Such backwashing with filtrate can delay the clogging of the membrane, thereby extending the remaining process time of the filter process. If, however, backwashing is carried out with another backwashing fluid, an interruption of the filtering process is generally necessary, for example in order to completely remove the backwashing fluid from the membrane filter before the filtering process is resumed. This is e.g. required for backwashing with oxygenated water in beer filtration.

Here and below, cleaning by backwashing is therefore to be understood as backwashing, in which the filtering process is interrupted, in particular the above-mentioned steps of retraction and prestressing may be necessary. When cleaning by backwashing, backwashing is (also) carried out with backwashing fluids other than the filtrate. Cleaning by means of backwashing can also comprise several backwashing steps, in particular with different backwashing media. The cleaning process can also be intensified by adapting the respective intensity and / or duration of individual backwashing steps. The type and / or amount of chemicals used can be adjusted to intensify a CIP cleaning process. For example, instead of the chemicals of an intermediate CIP, the chemical type, chemical quantity or concentration of a main CIP can be used. An intermediate CIP or intermediate cleaning differs from a main CIP or main cleaning in terms of duration, intensity and type, amount and concentration of the chemicals used.

In addition or as an alternative to the intensification of the cleaning process described, the cleaning process can be shortened. This can be done in particular in such a way that the second filtration unit for the first filtration unit at the end of the filtering process Filtering the fluid is available. In this way, an interruption in the production of the filter system can be avoided.

The influence of the at least one process parameter of the filtration with the first filtration unit on the regulation of filter processes or cleaning process in the second filtration unit can be unconditional, in the sense that the filtration with the second filtration unit is adjusted without direct reaction to the filtration with the first filtration unit becomes. The regulation of the filtration with the second filtration unit is therefore subordinate to the regulation of the filtration with the first filtration unit as the master. Alternatively, however, there can also be a reaction to the filtration with the first filtration unit. For example, separate control and / or regulating units of the first and second filtration units can be configured as slaves with respect to a higher-level regulating unit as masters. In this case, in addition to the intensification and / or shortening of the cleaning process of the second filtration unit, an extension of the filter process of the first filtration unit can be effected. As a result, in particular a compromise can be achieved between cleaning the second filtration unit as completely as possible and not excessively fouling the first filtration unit. This compensation can be effected in particular by the higher-level control unit mentioned above.

According to a further exemplary embodiment, a filter process of the second filtration unit can be extended depending on the process parameter of the filtration with the first filtration unit, in particular a predicted remaining process duration of a cleaning process of the first filtration unit. In this way it can be ensured that the cleaning of the first filtration unit is completed before the filtering process of the second filtration unit is ended. An extension of the filter process can, as mentioned above, for example, by increasing the limit value of the transmembrane pressure, by changing (depending on the filter state an increase or decrease can be useful) the crossflow volume flow and / or by lowering a pressure level of the filtrate, and / or by lowering the Filtrate volume flow and / or by changing the direction of the flow direction of the unfiltrate. Here too, the adaptation of the filter process of the second filtration unit can necessarily take place in dependence on the process parameter of the filtration with the first filtration unit or in retrospect on the control or regulation of the filtration of the first filtration unit.

In the above-mentioned exemplary embodiments, the filtering process or cleaning process of the second filtration unit and the cleaning process or filtering process of the first filtration unit run at least partially in parallel. This does not exclude that further process steps, for example the above-mentioned pretensioning or retraction of the membrane filter modules, are provided. These process steps can be integrated into the optimization of the occupancy control, whereby the duration of the further process steps can be predetermined or can also be controlled adaptively.

By alternately lengthening filter processes and / or intensifying and / or shortening cleaning processes of the first and second filtration units, the occupancy of the filtration units can be optimized. In particular, as mentioned above, standby phases or resting phases of individual filtration units can be largely avoided.

As already mentioned above, a prediction of a process duration or residual process duration of a filter process or cleaning process is often unreliable on the basis of models. According to a further development, the control unit and / or the separate control and / or control unit of the first filtration unit can therefore be designed to predict the process duration on the basis of fuzzy logic and / or artificial neural networks. A control or regulation based on fuzzy logic is able to lead unsharp data or parameters to a reliable statement. Such so-called expert systems do not require complete modeling of the filter system, but can rather be used as black box or gray box systems. In fuzzy logic, complex problems can easily be described by using fuzzy rules. The degree of belonging to the corresponding linguistic value is determined for each concrete input variable using the membership function of the fuzzy set (fuzzy set) of a linguistic value. The use of fuzzy logic to control and / or regulate a filter system enables the process control to be influenced by specifying simple and intuitive linguistic rules and values. A linguistic rule comprises a number of premises in the form of belonging to a number of input variables, i.e. Process parameters, to a number of linguistic values, which are linked by a logical link, the so-called precondition of the rule, and an action in the form of a membership function of an output variable, for example a control or regulation parameter and one of the above-mentioned actions, to one linguistic value.

According to this development, each rule can be specified by an expert, can also be learned by successive systems formed in cascades or by an automated method. In particular, an artificial neural network can be used in the automated method. Such an artificial neural network can be monitored, that is, logged and evaluated, of suitable process parameters Learn or adapt the rules of the filtration units again, whereby the observation can take place by an expert, in particular during the filtration operation.

A predetermined or learned rule can also be adapted through optimization steps. An optimization step can include the adaptation of a fuzzy set belonging to a linguistic value used in a rule or a prioritization or elimination of the rule. After evaluating the linguistic rules, the initial variable can be obtained by defuzzifying a membership function. By linking several linguistic rules through composition, even complex relationships in the area of process management can be easily formulated. In particular, composition can implement linguistic rules for regulating opposing trends, for example on the basis of the characteristic curve fields described in connection with FIG. 6, so that an optimization of the process control, in particular a reliable prognosis of the process duration, is possible on the basis of the fuzzy logic.

To use the fuzzy logic, the control unit and / or the separate control and / or regulating unit of the first filtration unit can in particular have a Mamdani controller or a Sugeno controller.

By alternative or additional use of an artificial neural network, the prognosis of the process duration can be trimmed towards an optimized expert system, which enables a reliable prognosis of the process duration or remaining process duration of filter processes and cleaning processes of the filtration units even without previous and external expert knowledge.

An artificial neural network consists of one or more artificial neurons, which are arranged in one or more layers. Each artificial neuron determines an output signal from one or more input signals. A net input can be determined from the one or more input signals with the aid of one or more predetermined weights as the sum of the weighted input signals. The output signal can be determined from the net input using an activation function. The activation function can be a threshold value function, a sigmoid function or a linear function. A sigmoid function has the advantage that it is continuously differentiable and can therefore be used in optimization processes such as the steepest gradient process. An artificial neuron can in particular be in the form of a variable threshold perceptron.

An artificial neural network has the particular advantage that it is an adaptive system. This involves learning an artificial neural network in the Generally, by adjusting the weights of the input signals to the neurons. For a multilayer neural network, such as the multilayer percepton (MLP), the back propagation algorithm can be used to carry out a learning step. An artificial neural network for predicting the process duration of filter processes or cleaning processes can be trained offline, ie without process control, by an expert, or (also) learn online (ie during ongoing process control).

In particular, an artificial neural network in the form of a neuro-fuzzy controller can be combined with a fuzzy controller. This allows the transparency of the intuitive rules of fuzzy systems to be combined with the learning ability of artificial neural networks. In particular, a neuro-fuzzy controller is able to learn linguistic rules and / or membership functions or to optimize existing ones.

A neuro-fuzzy controller can be implemented as a cooperative system in which the neural network operates independently of the fuzzy system, and the parameters of the fuzzy system are determined and / or optimized by the neural network. The learning of the neural network can be done by learning fuzzy sets or by learning linguistic rules. The learning of fuzzy sets can be done by a modified back propagation method, in which the position and form of the membership function of the fuzzy set is changed instead of the weights.

A neuro-fuzzy controller can also be implemented as a hybrid system in which the properties of fuzzy logic and the artificial neural network are inseparably combined. Alternatively or additionally, an artificial neural network can be linked to physical or process engineering models to form a hybrid controller. Instead of the weights, the fuzzy sets can take the place of a fuzzy neuron, whereby instead of the weighted sum and the activation function for the fuzzy neurons of an inner layer, the determination of the straight line of affiliation (fuzzyfication) for the input signals and their inference occurs. Instead of the weighted sum and the activation function for the fuzzy neurons of the starting layer, the composition and defuzzification can occur. One way of learning in the hybrid neuro-fuzzy controller is to specify all possible rules for the control and / or regulation of the filter system or filtration unit or the filtering process or cleaning process before commissioning the controller, and the neuro-fuzzy controller to regulate rules that are not required Have online operations eliminated.

One or more fuzzy controllers and / or artificial neural networks can be integrated into the control unit and / or the separate control and / or regulating unit of the first filtration unit in order to predict the process duration or remaining process duration. The linguistic rules of the fuzzy controller can include models, facts, environmental data, etc. and physically Show chemical relationships as expert knowledge. The artificial neural networks used can also be trained on the basis of such models or facts.

The use of fuzzy controllers and / or artificial neural networks to predict the process duration allows a significantly more reliable prognosis of the process duration to be made, even with inadequate modeling and unknown influences on the process duration. In particular, the system used can be designed to be capable of learning, so that the prognoses become better with increasing operating time of the filter system.

In addition or as an alternative to the above-described forecast of the process duration of filter processes and cleaning processes of the filtration units on the basis of fuzzy logic and / or artificial neural networks, the control unit can be designed to adaptively regulate the occupancy of the second filtration unit on the basis of a fuzzy Perform logic and / or artificial neural networks. For this purpose, the above-mentioned actions can be determined as output parameters of the fuzzy systems, artificial neural networks or neuro-fuzzy systems used, for example whether an intervention in an ongoing filter process or an ongoing cleaning process should take place. In particular, the termination of a filter process or the time of a filtration end can be determined as output parameters. Furthermore, the overflow speed, the filtration speed, the maximum transmembrane pressure, a point in time for a cleaning process by backwashing, a reversal of the unfiltrate flow, a point in time and a number of intermediate CIPs, and a point in time for a main CIP can be determined as starting parameters. Furthermore, a cleaning process can be initiated and / or the type and duration of the cleaning process can be selected, for example by backwashing, intermediate CIP or main CIP. Finally, the duration of a filter process can be adjusted, the duration of a soaking process of the membrane filter module can be adjusted, and a standby can be prepared.

The control unit and the separate control and / or regulating units of the filtration units can, in particular, as mentioned, be designed as programmable logic controllers, with a connection to the cloud for training the artificial neural networks being additionally provided. Data sets from other filter systems can thus be used to improve the adaptive control of the occupancy.

According to a further development, at least one sensor, in particular assigned to the first filtration unit, can be provided, the process parameter of the filtration with the first filtration unit comprising a process parameter of the filtration with the first filtration unit measured by means of the sensor. For example, a viscosity sensor and / or a density sensor can be provided which measures the viscosity of the unfiltrate or the density of the unfiltrate in the respective circuit of the first filtration unit. A sensor for measuring the viscosity can be designed as a quartz viscometer, in particular with piezomechanical quartz sensors. The density can be determined, for example, by radiation absorption, bending vibrators or the like. A sensor for measuring the viscoelasticity can also be provided. Since there is a connection between the viscosity of the unfiltrate and the yeast concentration (yeast cell number per milliliter) as well as the density and the yeast concentration, these parameters can be used to easily draw conclusions about the yeast cell number. Sensors for measuring the viscosity and / or the density can be arranged in situ in the filtration unit in a simple manner. For example, a Coriolis mass flow meter with induced torsional movement can be used to measure density and viscosity simultaneously. Further conceivable sensors for measuring process parameters include a sensor for measuring the turbidity of the unfiltrate, a sensor for measuring the color of the unfiltrate and / or the filtrate, a sensor for measuring the transmembrane pressure, a sensor for measuring the filtrate flow, and a sensor for measuring the Overflow speed, a sensor for measuring the mass flow through the membrane filter, a temperature sensor for the unfiltrate, a sensor for measuring a backwash volume flow, a sensor for measuring a backwash speed, a sensor for measuring a backwash pressure and the like.

The process parameter or parameters of the filtration with the first filtration unit measured by the sensors can be used as an alternative or in addition to the predicted process duration of a filter process and / or a cleaning process of the first filtration unit by the control unit according to the invention in order to adaptively control an occupancy of the second filtration unit. Process parameters relating to the filtration with the first filtration unit can thus be explicitly taken into account when regulating the occupancy of the second filtration unit. In this way, a more fundamental but more complex optimization of the occupancy of the second filtration unit can be achieved.

According to a development, the second filtration unit can comprise a plurality of membrane filter modules connected in parallel with a double inlet line and a double return line, controllable shut-off devices being provided between the membrane filter modules connected in parallel for reversing a direction of flow of the unfiltrate during the filtration, and the control unit being designed to reverse the flow direction of the unfiltrate in the second filtration unit as a function of the process parameter of the filtration with the first filtration unit. Each membrane filter module of the second filtration unit thus has two inlet lines which can be opened and closed by the control unit via controllable shut-off devices, for example shut-off valves, and which are connected to the inlet or the outlet of the membrane filter module for unfiltrate. At the same time, each membrane filter module has two return lines which can be selectively opened and closed by the control unit via controllable shut-off devices and which are also connected to the inlet or the outlet of the following membrane filter module for unfiltrate. Between two membrane filter modules, the feed lines of the subsequent module can be given by the return lines of the previous module. Controllable shut-off devices are thus provided between the membrane filter modules connected in parallel. By deliberately opening only one of the two shut-off devices in the feed lines or return lines, the individual membrane filter modules of the second filtration unit can reverse the direction of flow of the unfiltrate, even without reversing a pump that circulates the unfiltrate. The shut-off devices are always controlled so that the membrane filter modules are successively flowed through by the unfiltrate. By reversing the direction of flow of the unfiltrate, the duration of a filter process can be extended, since an area of the membrane filter modules that is initially less loaded by membrane fouling is now exposed to a higher load and vice versa. Such a reversal of the direction of flow of the unfiltrate can thus be used by the control unit in a targeted manner, for example depending on a remaining process duration of a cleaning process of the first filtration unit, in order to extend a filtering process of the second filtration unit, for example until the cleaning process of the first filtration unit is completed.

Alternatively, the second filtration unit can have membrane filter modules connected in series and a reversible pump with which the direction of flow of the unfiltrate can be reversed in a controlled manner.

The above-mentioned objects are also achieved by a method for filtering fluids, in particular with one of the devices described above, the method comprising a sequence of filter processes for filtering unfiltrate and of cleaning processes for cleaning the membrane filter modules of the first and second filtration units, wherein the unfiltrate is passed in separate circuits through at least one membrane filter module of the first filtration unit and through at least one membrane filter module of the second filtration unit, and wherein the filtrate is derived from the membrane filter modules, an occupancy of the second filtration unit depending on at least one process parameter of a filtration with the first filtration unit is controlled adaptively. The same variations and developments that were described above in connection with the devices for filtering fluids according to the invention can also be applied to the method for filtering fluids. In particular, the developments of the first and second filtration units, the membrane filter modules and the control and / or regulating units described above can be used. In addition to the filter processes and cleaning processes, the sequence can also include further sections, for example the above-mentioned pretensioning and retraction of the membrane filter modules, standby phases or rest phases and the like. As described above, the cleaning processes can in particular be backwashing processes, intermediate CIPs and main CIPs. The process parameters described above can be used as process parameters of the filtration with the first filtration unit.

Furthermore, as already mentioned above, in addition or as an alternative, an occupancy of the first filtration unit can be adaptively regulated as a function of at least one process parameter of a filtration with the second filtration unit. Again, the process parameters described above can be used to adaptively control the occupancy of the first filtration unit. In particular, the process parameter of the filtration with the first filtration unit can include a predicted process duration, in particular residual process duration, of a filter process and / or a cleaning process of the first filtration unit. Likewise, as mentioned above, the occupancy of at least one filtration unit of a plurality of filtration units can be adaptively regulated as a function of at least one process parameter of a filtration with one or more other filtration units of the plurality of filtration units. Furthermore, the occupancy of all filtration units of the large number of filtration units can be regulated adaptively.

The adaptive control of the occupancy of the second filtration unit can include adapting at least one control parameter of a filter process and / or a cleaning process of the second filtration unit, the control parameter of the filter process of the second filtration unit being adapted in particular as a function of a predicted remaining process duration of a parallel cleaning process of the first filtration unit can, and / or wherein the control parameter of the cleaning process of the second filtration unit can be adapted in particular as a function of a predicted remaining process duration of a parallel filter process of the first filtration unit. The filtering process and the cleaning process do not have to run completely parallel. The control parameters described above can be used as control parameters. In particular, limit values for process parameters such as the transmembrane pressure, in which a specific action, for example ending a filter process and initiating cleaning or backflushing with filtrate to extend the filtering process, can be carried out as control parameters.

For example, if the second filtration unit is in a filtering process while the first filtration unit is in a cleaning process, the adaptive control can extend the filtering process at least until the cleaning process of the first filtration unit is complete. In addition, the filter process can be extended until the pre-tensioning of the membrane filter modules of the first filtration unit that follows the cleaning process is completed. If, on the other hand, the second filtration unit is in a cleaning process while the first filtration unit is in a filter process, the adaptive control can intensify and / or shorten the cleaning process as described above, so that the second filtration unit is available to the first filtration unit at the end of the filtering process .

The adaptive control of the occupancy of the second filtration unit can include stopping a filter process and initiating a cleaning process of the second filtration unit and / or terminating the cleaning process and initiating a further filter process of the second filtration unit depending on the process parameter of the filtration with the first filtration unit. By adaptively ending a filter process and initiating a cleaning process of the second filtration unit in dependence on the process parameter of the filtration with the first filtration unit, it is possible to react flexibly to an early availability of the first filtration unit, so that the effectiveness of cleaning the second filtration unit can be improved. By adaptively ending the cleaning process and initiating a further filter process of the second filtration unit in dependence on the process parameter of the filtration with the first filtration unit, it is possible to react flexibly to a premature end of a filter process of the first filtration unit. In this way, an interruption in the filtration of the entire filter system can be avoided.

The initiation of the cleaning process of the second filtration unit can in particular include a selection of the cleaning process from the following group: cleaning by backwashing, cleaning-in-place, cleaning by backwashing and cleaning-in-place, general cleaning and predetermined cleaning programs. For example, a cleaning process by backwashing as described above can comprise several backwashing steps, in particular with different backwashing media. As described above, the initiation of the cleaning process can also include a selection of the type and amount of chemicals used in a CIP cleaning and / or the intensity and / or duration of the cleaning process and / or the adjustment of control parameters of the cleaning process. For example, an intermediate CIP with a small amount or concentration or a main CIP with a higher quantity or concentration and possibly other or additional chemicals. In the case of a general cleaning, individual or all filter membranes can even be replaced. In addition to the at least one process parameter of the filtration with the first filtration unit, at least one process parameter of a previous filter process of the second filtration unit can also be taken into account. As a result, the subsequent cleaning process can be adapted such that a fouling layer on the membrane filter modules of the second filtration unit is removed as completely as possible, taking into account the boundary conditions with regard to the occupation of the second filtration unit, in order to ensure a maximum process duration of the subsequent filter process.

According to a development, the adaptation of the at least one control parameter of the filter process of the second filtration unit, as described above, can include a reversal of a flow direction of the unfiltrate in the at least one membrane filter module of the second filtration unit. The adaptation can in particular include the determination of a point in time at which the flow direction is reversed. In this way, as mentioned above, a remaining process time of the filtering process of the second filtration unit can be extended. The reversal of the flow direction of the unfiltrate can be brought about by targeted activation of the shut-off devices described above, for example shut-off valves, in the inlet and return lines of the membrane filter modules of the second filtration unit connected in parallel.

As described above, the adaptive control of the occupancy of the second filtration unit can be carried out on the basis of fuzzy logic and / or artificial neural networks. Furthermore, as described above, the process duration of a filter process or a cleaning process can be predicted on the basis of fuzzy logic and / or artificial neural networks. Finally, the control and / or regulation of a filter process or a cleaning process of each filtration unit can also be carried out by a corresponding separate control and / or regulating unit of this filtration unit on the basis of fuzzy logic and / or artificial neural networks. In this way, by means of simple, intuitive rules and by training the artificial neural networks, reliable forecasts of the process duration or remaining process duration of filter processes and cleaning processes can be made, which can be used according to the method according to the invention for optimizing the occupancy of the filtration units.

In the filtration of beer in particular, there is a strong, difficult to predict dependence of the service life of membrane filter modules used on the ones used Raw materials. In spite of the difficulty in predicting a remaining remaining process time of a filter process, the devices and methods according to the invention permit an optimization of the occupancy of the filtration units of a filter system with a view to avoiding production interruptions and downtimes of individual filtration units. Both the type and duration of the cleaning processes used can be optimized, and the length and periods of the filter processes of different filtration units can be coordinated. As a result, the excessive use of cleaning agents, ie chemicals, can be avoided in CIP cleaning, since the type and duration of the cleaning processes used are adaptively regulated as required. Excessive clogging of the membrane filter can also be avoided by uncontrollably increasing the transmembrane pressure.

The devices and methods according to the invention achieve an occupancy and overall process optimization which lead to optimally utilized filtration units with optimal cleaning and efficient total operating costs. In contrast to the otherwise usual rigid processes of the filter processes and cleaning processes, the occupancy of the filtration units is now flexibly and adaptively adjusted by filter processes and cleaning processes in order to optimize the overall occupancy of the filter system. This affects not only the length of a standby phase, but all process steps starting with soaking, i.e. the soaking with chemicals during CIP cleaning, the CIP duration and the length of the filtering process. Since fewer chemicals can be used in CIP cleaning, the service life of the filter membranes of the membrane filter modules is also extended, which means that additional costs can be saved.

Filtration of beer generally results in shorter cycles from the filtering process and cleaning process compared to other fluids, so that optimized occupancy control in breweries is particularly important. For breweries with continuous operation, the optimized occupancy control described above leads to more efficient filtration and thus to cost savings. Smaller filter systems can produce a higher amount of filtrate, which reduces the total cost of ownership. In addition, the filter membranes are protected.

Further features and exemplary embodiments and advantages of the present invention are explained in more detail below with reference to the drawings. It is to be understood that the embodiments are not exhaustive of the scope of the present invention. It goes without saying that some or all of the features described below can also be combined with one another in other ways.

Figure 1 shows a rough schematic of the structure of a single filtration unit. Figure 2 shows roughly schematically a filter system with several filtration units according to the present invention.

FIG. 3 shows the exchange of process parameters between separate control and / or regulating units of two filtration units according to the present invention.

FIG. 4 shows an alternative development for exchanging process parameters of a higher-level control unit.

FIG. 5 shows an example of an occupancy optimization for two filtration units.

FIG. 6 schematically shows the reversal of the flow direction in a filtration unit with several membrane filter modules connected in parallel.

FIG. 7 schematically shows a large number of characteristic curve fields for controlling and / or regulating the filtration units.

In the figures described below, the same reference symbols designate the same elements. For better clarity, the same elements are only described when they first appear. However, it goes without saying that the variants and embodiments of an element described with reference to one of the figures can also be applied to the corresponding elements in the other figures.

FIG. 1 shows, roughly schematically, an exemplary structure of an individual, independently controllable and / or regulatable filtration unit. The filtration unit shown here carries out a filtration using the crossflow method. However, it goes without saying that the further developments described below can also be used for filtration using the dead-end flow method by simple adjustments.

The filtration unit shown in FIG. 1 comprises at least one membrane filter module 1, for example a crossflow membrane filter module, through which the unfiltrate, for example unfiltered beer, can be circulated (indicated by arrow K). Of course, several membrane filter modules can also be provided, which can be arranged in series or parallel to one another. A special variant of this is shown in FIG. 6, for example. In crossflow filtration, the unfiltrate is flowed along the membrane of the membrane filter modules and emerges vertically from the membrane filter as filtrate. For this purpose, a filtrate line 18 is provided with a control or regulating valve 17, via which the filtrate is withdrawn from the membrane filter module 1. The control valve 17 can be controlled or regulated by the control and / or regulating unit 8 of the filtration unit, for example in order to set a pressure level on the filtrate side. The position of the control valve 17 is thus a manipulated variable of the control and / or regulation of a filter process of the filtration unit, for example the duration of the filter process can be controlled or regulated. Correspondingly, for example, the pressure level on the filtrate side is a control or regulating parameter of the control and / or regulating unit 8.

The filtration unit also has an inlet line 3, through which, for example, beer loaded with yeast is pumped in the direction of the membrane filter module 1, for example via a, in particular controllable, pump 7. Unlike in FIG. 1, the pump 7 can also be arranged on the section of the feed line 3 between the confluence of the line 13 and the point 16. Furthermore, the filtration unit comprises a circulation line 5, into which the inlet line 3 opens at point 16, and through which the unfiltrate, i.e. the retentate can circulate in the circuit K. Finally, the filtration unit also has a return line 6, via which unfiltrate can be discharged from the circulation line 5. For this purpose, a control valve 9 is provided in the return line 6, via which the flow of the derived unfiltrate can be adjusted. Furthermore, a control valve 19 is provided in the circulation line 5, via which the flow of the circulating unfiltrate can be adjusted. Additionally or alternatively, a, in particular controllable, pump 15 can be provided in the circulation line 5, by means of which, for example, an overflow speed of the filter membrane through the non-filtrate can be set. The transmembrane pressure can be reduced, for example, by changing (depending on the filter state, an increase or decrease) of the overflow speed. A drain line 12 branches off from the return line 6, via which the unfiltrate from the return line 6 can be fed to a channel 10 via a further control valve 11. The flow of the unfiltrate to this channel can be set via the control valve 11. The return line 6 opens into a concentration tank 2, which is designed here, for example, as a cylindroconical tank and which is connected to the inlet line 3 via a line 13 and a control valve 14. Concentrated unfiltrate can thus be returned to the feed line 3 from the concentration tank 2.

Furthermore, in the non-limiting further development of FIG. 1, sensors 4a and 4b are arranged in the circulation line 5 and the return line 6, respectively. Using these sensors, process parameters, such as the viscosity or density of the unfiltrate, can be measured in situ during the filtration. As described in detail above, a viscosity sensor and / or a density sensor can be used as sensors, for example. The measured process parameters are transmitted from the sensors 4a and 4b to the control and / or regulating unit 8 via lines (not shown) or wirelessly, where they are processed with the filtration unit to control and / or regulate the filtration. It goes without saying that the specific arrangement of the sensors 4a and 4b in FIG. 1 is not limiting, but is only illustrative. Fewer or more sensors can be provided, which can be arranged at different locations on the filtration unit. The sensors can also send the measured process parameters to a transmit higher-level control unit, which it uses to control the occupancy of at least one further filtration unit.

The control valves 9, 11, 14, 17 and 19 as well as the pumps 7 and 15 can be used by the control and / or regulating unit 8 to control or regulate the filtration. For example, by controlling the pump 15, a flow rate or overflow rate of the unfiltrate in the membrane filter module 1 can be set as mentioned above. By controlling the pump 7, the pressure level on the unfiltrate side can be influenced. By controlling the control valve 17, however, the pressure level on the filtrate side can be influenced. By actuating the pump 7 and / or actuating the control valve 17, the trans-membrane pressure and its gradient can thus be influenced directly via the filter membrane of the membrane filter module 1. In the filtration unit shown, the control and / or regulating unit 8 takes over the control or regulation of the above-mentioned control or regulating valves and pumps via corresponding control or regulating parameters. However, it is also conceivable that a higher-level control unit controls individual or all of the above-mentioned control valves and pumps directly, for example to extend a filtering process.

The filtration unit can also have cleaning devices, not shown, such as a backwash line for a cleaning process by means of backwashing, and a CIP unit. The cleaning devices can also be controlled or regulated by the control and / or regulating unit 8 and / or directly by a higher-level regulating unit. For example, a higher-level control unit, as described above, can intensify and / or shorten a cleaning process, for example by controlling a controllable pump used for backwashing.

As described in detail above, the control and / or regulation of filter processes and cleaning processes of the filtration unit can be carried out by means of the control and / or regulating unit 8 on the basis of fuzzy logic and / or artificial neural networks. Intuitive linguistic rules can be used, which sometimes combine opposing criteria with one another in order to obtain an optimal filtration result. As described above, the individual rules can be weighted differently and / or prioritized and / or eliminated. The linguistic rules can be derived, for example, from so-called characteristic curve fields, as shown in FIG. 7. Some exemplary, non-limiting characteristic curve fields for the control and / or regulation of a filtration unit are shown in FIG.

The sub-figure 7a) shows, for example, a characteristic curve of sustainability in relation to the output of filtrate. The sub-figure 7b) shows the characteristic curve of sustainability with reference to an energy input. The sub-figure 7c) shows a characteristic of the beer quality with reference to one Cleaning effort. Sub-figure 7d), on the other hand, shows a characteristic curve of the energy input with reference to the output. The sub-figure 7e) shows a characteristic curve of the discharge with reference to the membrane lifetime of the filter membrane. Finally, sub-figure 7f) shows the characteristic curve of sustainability with reference to the cleaning time.

For example, sustainability can be increased by using less media (a lower concentration of chemicals, or omitting a CIP component, for example enzymes) when cleaning the filter membranes, which means that cleaning takes longer or is not as effective. In the subsequent filtering process, less filtering (for example, less beer is filtered) or only a short filtering process can be achieved before the filter membrane has to be cleaned again.

The use of energy can be reduced by saving energy, for example by CIP cleaning at a lower temperature. This also means that cleaning takes longer or is not as effective, as a result of which the output quantity is reduced and / or the process duration of the filter process is shortened.

In general, a high amount of filtrate is the primary goal when operating a filter system without considering sustainability and membrane life. For this purpose, CIP cleaning can be carried out faster and with higher concentrations, temperatures and / or greater media use in order to reduce downtimes, i.e. To minimize times without filtration of the filtration unit. However, this generally shortens the membrane life as shown in sub-figure 7e).

The beer quality is above all other sizes and must be met, especially for premium varieties. If necessary, the filtration speed is reduced for this. A CIP cleaning can be carried out longer and more thoroughly, whereby the CIP cleaning may be carried out with high temperatures and high cleaning agent concentrations or additional additives. Often, beers that are fermented too short and have too high a yeast cell count are not filtered. In general, moderate operating conditions, e.g. moderate transmembrane pressures, the beer quality of the filtrate.

To improve the sustainability of the filtration, cleaning can only be carried out as required. Ie it is only cleaned as much and only as and when it is necessary. An evaluation of the cleaning success by appropriately arranged sensors, for example for measuring turbidity in the backwashing fluid, can have a supporting effect. Since particles released or detached from the membrane, such as, for example, yeast, cause turbidity, the success and duration of backwashing can be assessed. An example of the optimization of sustainability with demand-driven cleaning is shown in sub-figure 7f). In order to protect the filter membranes from membrane breakage, the transmembrane pressures must not be too high. On the one hand, this limits the filtration speed. On the other hand, the filter membranes can be protected by moderate cleaning, ie low temperatures and low concentrations of chemicals. In order to guarantee adequate cleaning of the filter membranes, the process duration of the filter process must be adjusted accordingly. In addition, moderate cleaning takes longer, which also extends the process duration of the cleaning process. In return, however, the membrane life increases, so that costs can be reduced by the membrane exchange. However, this also reduces the amount of filtration.

The exemplary rules described above partially set opposite trends for the control or regulation of the filtration unit. With the help of fuzzy logic and / or artificial neural networks, however, an optimal balance between these trends can be determined, since no detailed modeling of the filtration or cleaning is required.

Figure 2 shows roughly schematically a filter system with several filtration units according to the present invention. Each of the illustrated, in principle independently controllable and / or regulatable, filtration units 120-1 to 120-N comprises at least the circuit K of the filtration unit shown in FIG. 1 with at least one membrane filter module 1, the circulation line 5, the pump 15, the control valve 19 and for example a sensor 4a. However, modifications of this circuit can be used. For example, each filtration unit can have its own separate control and / or regulating unit 108-1 to 108-N, which controls and / or regulates the filtration with the respective filtration unit. Likewise, each filtration unit 120-1 to 120-N can have its own concentration tank 2, its own return line 6 with control valve 9 and possibly sensor 4b and its own drain line 12 to the channel 10 as shown in FIG. 1.

However, to illustrate the present invention of the adaptive occupancy control, the filtration units 120-1 to 120-N are shown in simplified form, only the separate control and / or regulating units 108-1 to 108-N being shown. A return line with control valve 109-1 to 109-N leads from each filtration unit to a common return line 106 for the filtration units, which leads the return of unfiltered material to a common concentration tank 102. In this common return line 106 there is in turn a sensor 104 which can measure, for example, a viscosity of the returning unfiltrate. Even if not shown, a sensor can also be located in each return line of each filtration unit 120-1 to 120-N. Likewise, a common outlet line 1 12 with control valve 1 1 1 to a channel 1 10 is provided for unfiltrate to be discarded. A common concentrate tank 102 can provide, for example, a balance in the process duration of filter processes with different filtration units. From the concentration tank 102, a line 11 leads again via a control valve 114 into a common inlet line 103 for unfiltered fluid, in particular beer, which is supplied by a controllable pump 107 from an upstream system part, for example a storage tank. Here, too, the pump 107 can alternatively be arranged between the confluence of the line 13 and the branch to the individual filtration units. Furthermore, in the non-limiting further development shown here, a sensor 124 is provided, for example for measuring a yeast cell number of the supplied unfiltrate.

The unfiltrate supplied via the common inlet line 103 is led via branched inlet lines to the filtration units 120-1 to 120-N, separate control valves 123-1 to 123-N being provided in the individual inlet lines, by means of which the volume flow of unfiltrate to the respective filtration unit can be controlled or regulated. Controlled opening and closing of the control valves 123-1 to 123-N can ensure that only those filtration units that are currently in the status of the filtering process are supplied with unfiltrate. In contrast, the corresponding control valve can be closed during a cleaning process of a filtration unit.

Separate filtrate lines with their own control valves 1 17-1 to 1 17-N lead from the filtration units 120-1 to 120-N to a common filtrate line 1 18, which forwards the filtrate to a downstream system part, for example a pressure tank or a filling system.

FIG. 2 also shows a higher-level control unit 180, by means of which the assignment of the filtration units 120-1 to 120-N is adaptively controlled, as described above. For this purpose, the control unit 180 can either directly or indirectly control valves 123-1 to 123-N, 1 17-1 to 1 17-N and 109-1 to 109-N, as well as separate circuit pumps 15 (not shown in Figure 2) of the respective Control or regulate filtration units. In the case of indirect regulation of the occupancy of the filtration units, the control unit 180 passes on corresponding control signals to the respective separate control and / or regulating units 108-1 to 108-N, which are transmitted by the separate control and / or regulating units in the control or regulation of the Filter processes and cleaning processes of the respective filtration units are taken into account. Such control signals can signal, for example, a desired extension of a filter process, a desired intensification and / or shortening of a cleaning process or the like, as described in detail above. In the case of direct control, the control unit 180 can also send control signals directly to the control valves and pumps mentioned above. The control of the occupancy of a filtration unit by means of the control unit 180 is carried out as described in FIG Dependence on at least one process parameter of a filtration with at least one further filtration unit from the group of filtration units 120-1 to 120-N. The control unit 180 adapts filter processes and cleaning processes of the filtration units 120-1 to 120-N as described below using the example in FIG. 5 in order to optimize the occupancy of the entire filter system.

Instead of the higher-level control unit 180, the control of the occupancy of each filtration unit can, however, as described above, also be carried out by the separate control and / or regulating units 108-1 to 108-N. An example of such an individual occupancy control is demonstrated in FIG. 3 using the example of two filtration units. For the sake of simplicity, only the control and / or regulating units 208-1 and 208-2 and, by way of example, one sensor 204-1 and 204-2 are shown for each of the filtration units.

The sensor 204-1 of the first filtration unit transmits a measured process parameter 226-1 of a filtration with the first filtration unit to the control and / or regulating unit 208-2 of the second filtration unit. Conversely, the sensor 204-2 of the second filtration unit transmits a measured process parameter 226-2 of a filtration with the second filtration unit to the control and / or regulating unit 208-1 of the first filtration unit. Furthermore, the control and / or regulating unit 208-1 of the first filtration unit predicts a process duration, in particular the remaining process duration, a filter process or a cleaning process of the first filtration unit, as described above in detail, and transmits the predicted process duration 225-1 to the control and / or Control unit 208-2 of the second filtration unit. Correspondingly, the control and / or regulating unit 208-2 of the second filtration unit predicts a process duration, in particular remaining process duration, a filter process or a cleaning process of the second filtration unit and transmits the predicted process duration 225-2 to the control and / or regulating unit 208-1 of the first Filtration unit.

Depending on the transmitted process parameter 226-2 and / or the transmitted process duration 225-2, the control and / or regulating unit 208-1 of the first filtration unit regulates an assignment of the first filtration unit as described above, for example by extending a filter process or intensifying and / or or shortening a cleaning process of the first filtration unit. Correspondingly, the control and / or regulating unit 208-2 of the second filtration unit controls an occupancy of the second filtration unit as a function of the transmitted process parameter 226-1 and / or the transmitted process duration 225-1. As described above, fuzzy logic and / or artificial neural networks can be used. In the development of FIG. 3, the control and / or regulating units 208-1 and 208-2 of the filtration units are configured to be equivalent. However, a control and / or regulating unit can also be configured as a slave of the other control and / or regulating unit as a master. This is particularly conceivable if only the occupancy of one filtration unit from this pair is to be regulated as a function of at least one process parameter of a filtration with the other filtration unit.

In particular if a large number of filtration units are provided, it is advisable to regulate the occupancy of the filtration units by means of a higher-level control unit. This situation is shown in simplified form in FIG. 4, a higher-level control unit 380 being configured as the master compared to the separate control and / or control units 308-1 and 308-2 of the filtration units configured as slaves.

In the architecture of FIG. 4, the control and / or regulating unit 308-1 of the first filtration unit transmits a predicted process duration 325-1 to the higher-level regulating unit 380. Furthermore, a sensor 304-1 shown as an example transmits a measured process parameter 326- to the first filtration unit. 1 to the control unit 380. The transmitted process parameter 326-1 and / or the transmitted process duration 325-1 are processed by the control unit 380 in order to send a control signal 327-2 to the control and / or regulating unit 308 for regulating the occupancy of the second filtration unit -2 to transmit this second filtration unit. Depending on the control signal 327-2, the control and / or regulating unit 308-2 of the second filtration unit can then adapt a filter process or a cleaning process of the second filtration unit accordingly, as described above.

Conversely, the control and / or regulating unit 308-2 of the second filtration unit transmits a process duration 325-2 to the higher-level control unit 380. Likewise, a sensor 304-2 of the second filtration unit transmits a measured process parameter 326-2 to the control unit 380. The transmitted process duration 325-2 and / or the transmitted process parameter 326-2 are processed by the control unit 380 in order to transmit a control signal 327-1 for controlling the occupancy of the first filtration unit to the control and / or regulating unit 308-1 of this first filtration unit. Depending on the control signal 327-1, the control and / or regulating unit 308-1 of the first filtration unit can then adapt a filter process or a cleaning process of the first filtration unit accordingly, as described above.

Mixed forms of the architectures in FIGS. 3 and 4 are also conceivable in a filter system with several filtration units. In FIG. 5, the occupancy control according to the invention is demonstrated using an example occupancy optimization for two filtration units. The assignments for the first filtration unit A and the second filtration unit B are shown here by way of example using sequences with the steps of pre-tensioning, filtration, shutdown, CIP and pause. However, it goes without saying that the present invention is not limited to these special steps, but alternative or additional cleaning processes and further process steps can be provided. It goes without saying that the occupancy optimization can be expanded accordingly to three or more filtration units. For example, an occupancy rule can guarantee that at least one filtration unit of a plurality of filtration units is always in a filter process. If the number of filtration units is sufficiently high, for example 10 or more, the control can also cause the output of filtrate to remain approximately constant. In addition to the occupancy control, suitable control parameters of the filter processes, for example the transmembrane pressure and / or the overflow speed, can be adapted to adapt the instantaneous output to the filtrate of the individual filter processes in such a way that the total output is approximately constant.

The uppermost pair A1 / B1 in FIG. 5 shows, as is known in the prior art, rigid sequence sequences for the individual process steps. In the sequences shown, it is assumed for demonstration purposes that the filtration with the two filtration units takes place under exactly the same circumstances, so that an optimal occupation without filtration gaps, in particular breaks, apart from an initial pause of the second filtration unit, is possible. In this idealized situation, during the filtering process of the second filtration unit B1, the sequence of shutting down, CIP cleaning and pretensioning takes place in the first filtration unit A1, so that the first filtration unit A1 is again available to the second filtration unit B1 at the end of the filtering process. The same applies in reverse for the second filtration unit B1.

In reality, however, the filtrations with the individual filtration units never run completely the same, so that, as indicated in the middle pair A2 / B2, there are inevitably shifts and associated filtration gaps in the sequences. For example, the second filtration of the first filtration unit A2 in the sequence illustrated by way of example extends beyond the pretensioning of the second filtration unit B2, so that there is a shift here. Other events can also lead to filtration gaps or breaks in the sequence. For example, a threshold value for the maximum concentration of the unfiltrate of a filtration unit can lead to an unexpectedly premature end of a filtering process of this filtration unit, so that no other filtration unit is available at this time. Since the sequence sequences in the prior art are rigidly specified, for example in the form of fixed limit values are specified, a once-occurring shift continues through the operation of the filter system, without the occurring filtration gaps being able to be corrected.

With the present invention of the adaptive control of an occupancy of the filtration units, the occurrence of filtration gaps can be effectively avoided. The adaptive control is shown as an example in FIG. 5 using the last pair A3 / B3.

In the sequence shown, it is assumed as an example that the filtration of the first filtration unit A3 is controlled or regulated autonomously by the control and / or regulating unit of this filtration unit, i.e. regardless of the filtration of the second filtration unit B3. Conversely, the occupancy of the second filtration unit, i.e. In particular, the times and duration of the filter processes and the cleaning processes of the second filtration unit B3, adaptively regulated as a function of predicted remaining process times of filter processes and cleaning processes of the first filtration unit A3.

Without restriction, the sequence of the first filtration unit A3 begins with a particularly long filter process, so that the first filter process of the second filtration unit B3 starts with an adaptively controlled delay. Since the first filter process of the illustrated sequence of the first filtration unit A3 is longer than usual, the higher-level control unit or the separate control and / or regulating unit of the first filtration unit A3 predicts an extended process duration of the subsequent CIP cleaning. In order to avoid the occurrence of a filtration gap due to this longer CIP cleaning, the first filtering process of the second filtration unit B3 is extended until the CIP cleaning and the subsequent pretensioning of the first filtration unit A3 are completed. In this way it is ensured that the first filtration unit A3 is again available for a subsequent filtering process at the end of the filtering process of the second filtration unit B3.

In the exemplary sequence shown, it is now assumed that the second filtering process of the first filtration unit A3 must be ended early because the filter membrane clogs up more quickly than expected. On the basis of one or more measured process parameters of this filter process, the control unit or the separate control and / or regulating unit of the first filtration unit A3 predicts this shortened process duration of the filter process. In order to prevent the occurrence of a filtration gap, the CIP cleaning of the second filtration unit B3 is therefore shortened, so that the second filtration unit B3 is again available for filtering product such as beer at the end of the shortened process duration of the filtering process of the first filtration unit A3. In order to guarantee an effective cleaning of the filter membrane heavily clogged by the previous, extended filter process, the control unit additionally intensifies the CIP cleaning of the second filtration unit B3. In this way, the control unit can counteract instability in the sequence of filter processes and cleaning processes.

As exemplified in the example in FIG. 5, an adaptive control of the occupancy of one or more filtration units leads to an optimal utilization of the filter system, whereby the filter membranes can be protected and the use of chemicals can be reduced.

As mentioned above, a filter process can be extended, for example, by reversing the direction of flow of unfiltrate in a filtration unit with several membrane filter modules connected in parallel or in series. Such a reversal of the flow direction is shown as an example in FIG. 6.

For the sake of simplicity, only the elements of the filtration unit required for reversing the flow direction are shown in FIG. 6. As mentioned, the filtration unit has a plurality of membrane filter modules 401 connected in parallel, which are connected to one another and to a common inlet line 435 and a common return line 436 via double inlet lines 431 and 432 and double return lines 433 and 434. By way of example, a pump 415 for circulating the unfiltrate is provided in the common feed line 435.

Controllable shut-off valves 430 are provided between the membrane filter modules 401 connected in parallel and are specifically opened or closed by a control and / or regulating unit of the filtration unit. In Figure 6, closed check valves are filled and open check valves are shown empty. In addition to the check valves between adjacent membrane filter modules 401, check valves 430 are also provided in the feed lines 431 and 432 of the first membrane filter module.

The shut-off valves 430 are opened or closed by the control and / or regulating unit in such a way that, as shown by arrows in FIG. 6, there is a flow through the membrane filter modules 401. Alternating opening and closing of the shut-off valves 430 in the feed lines results in a flow of the unfiltrate in the form of an S-curve, as shown in the upper figure in FIG. 6, with direct successive membrane filter modules 401 always being flowed through in the opposite direction.

If the configuration of the check valves 430 of the upper part figure is inverted, the flow direction of the unfiltrate through the membrane filter modules 401 is exactly reversed as shown in the lower part figure. Due to the design of the filtration unit, this is possible without reversing the pumping direction of the pump 415. As a result, the roles of inlet and outlet for the unfiltrate are reversed on all membrane filter modules 401. Thereby For example, a thickness of the fouling layer along the filter membranes of membrane filter modules 401 can be compensated, which can lengthen the filtering process.

If the membrane filter modules are arranged in series, a reversal of the flow direction of the unfiltrate can be achieved, as mentioned above, by reversing a reversible pump.

As described above, the reversal of the flow direction of the unfiltrate through the membrane filter modules 401 can be forced by the higher-level control unit as part of the adaptive occupancy control.

The further developments described for adaptive occupancy control of filtration units of a filter system allow effective avoidance of filtration gaps and at the same time optimal cleaning of the membrane filter modules. On the one hand, this can extend the life of the membrane, and on the other hand, chemicals and energy can be saved, which can improve the sustainability of filtration. By optimizing the occupancy plan, the output of the filter system can also be increased, which improves the economy of the system.

Claims

Expectations

A device for filtering fluids, in particular beer, comprising: a first, independently controllable and / or regulatable filtration unit (120-1) with at least one membrane filter module (1, 401); a second, independently controllable and / or regulatable filtration unit (120-2) with at least one membrane filter module (1, 401); and at least one control unit (180, 108-2, 208-2, 380); characterized in that the control unit (180, 108-2, 208-2, 380) is designed to occupy the second filtration unit (120-2) as a function of at least one process parameter (225-1, 226-1, 325- 1, 326-1) adaptively regulate a filtration with the first filtration unit (120-1) and / or occupancy of the first filtration unit (120-1) as a function of at least one process parameter (225-2, 226-2, 325- 2, 326-2) adaptively regulate a filtration with the second filtration unit (120-2).

2. Device according to claim 1, wherein the process parameter (225-1, 226-1, 325-1, 326-1) of the filtration with the first filtration unit (120-1) one by means of the control unit (180, 108-

2. 208-2, 380) or a separate control and / or regulating unit (8, 108-1, 208-1, 308-1) of the first filtration unit predicted process time (225-1, 325-1), in particular remaining process time, a filter process and / or a cleaning process of the first filtration unit (120-1).

3. Device according to claim 1 or 2, wherein the control unit (180, 108-2, 208-2, 380) is designed to: a filtering process of the second filtration unit (120-2) depending on the process parameter (225-1, 226 -1, 325-1, 326-1) of the filtration with the first filtration unit (120-1), in particular a predicted remaining process time (225-1, 325-1) of a cleaning process of the first filtration unit (120-1); and / or a cleaning process of the second filtration unit (120-2) depending on the process parameter (225-1, 226-1, 325-1, 326-1) of the filtration with the first filtration unit (120-1), in particular a predicted one Remaining process duration (225-1, 325-1) of a filter process of the first filtration unit (120-1), to be intensified and / or shortened.

4. The device according to claim 2 or 3, wherein the control unit (180, 108-2, 208-2, 380) and / or the separate control and / or regulating unit (8, 108-1, 208-1, 308-1 ) the first Filtration unit (120-1) is designed to predict the process duration on the basis of fuzzy logic and / or artificial neural networks.

5. Device according to one of the preceding claims, wherein the control unit (180, 108-2, 208-2, 380) is designed to adaptively control the occupancy of the second filtration unit (120-2) on the basis of fuzzy logic and / or perform artificial neural networks.

6. Device according to one of the preceding claims, further comprising at least one sensor (4a, 4b, 204-1, 304-1), in particular assigned to the first filtration unit (120-1), the process parameter (225-1, 226- 1, 325-1, 326-1) of the filtration with the first filtration unit (120-1) comprises a process parameter (226-1, 326-1) of the filtration with the first filtration unit (120-1) measured by means of the sensor.

7. Device according to one of the preceding claims, wherein the second filtration unit (120-2) comprises a plurality of membrane filter modules (401) connected in parallel with a double feed line (431, 432) and a double return line (433, 434), between the parallel ones Membrane filter modules (401) controllable shut-off devices (430) are provided for reversing a flow direction of the unfiltrate during the filtration, and wherein the control unit (180, 108-2, 208-2, 380) is designed to determine the flow direction of the unfiltrate in the second Reverse filtration unit (120-2) depending on the process parameter (225-1, 226-1, 325-1, 326-1) of the filtration with the first filtration unit (120-1).

8. A method for the filtration of fluids, in particular with a device according to one of claims 1 to 7, comprising a sequence of:

Filter processes for filtering unfiltrate, the unfiltrate being passed in separate circuits through at least one membrane filter module (1, 401) of a first filtration unit (120-1) and through at least one membrane filter module (1, 401) of a second filtration unit (120-2), and the filtrate is derived from the membrane filter modules (1, 401); and

Cleaning processes for cleaning the membrane filter modules (1, 401) of the first and second filtration units (120-1, 120-2); characterized in that an assignment of the second filtration unit (120-2) is adaptively controlled as a function of at least one process parameter (225-1, 226-1, 325-1, 326-1) of a filtration with the first filtration unit (120-1).

9. The method according to claim 8, wherein an additional occupation of the first filtration unit (120-1) as a function of at least one process parameter (225-2, 226-2, 325-2, 326-2) of a filtration with the second filtration unit (120 -2) is regulated adaptively.

10. The method according to claim 8 or 9, wherein the process parameter (225-1, 226-1, 325-1, 326-1) of the filtration with the first filtration unit (120-1) a predicted process time (225-1, 325- 1), in particular the remaining process time, a filter process and / or a cleaning process of the first filtration unit (120-1).

1 1. Method according to one of claims 8 to 10, wherein the adaptive control of the occupancy of the second filtration unit (120-2) comprises adapting at least one control parameter of a filter process and / or a cleaning process of the second filtration unit (120-2), the control parameter of the filter process the second filtration unit (120-2) is adapted in particular as a function of a predicted remaining process duration (225-1, 325-1) of a parallel cleaning process of the first filtration unit (120-1), and / or wherein the control parameter of the cleaning process of the second filtration unit (120-2) especially depending on a predicted remaining process time (225-1,

325-1) of a parallel filter process of the first filtration unit (120-1) is adapted.

12. The method according to any one of claims 8 to 1 1, wherein the adaptive control of the occupancy of the second filtration unit (120-2) ending a filtering process and initiating a cleaning process of the second filtration unit (120-2) and / or ending the cleaning process and Initiation of a further filter process of the second filtration unit (120-2) depending on the process parameter (225-1, 226-1, 325-1,

326-1) of filtration with the first filtration unit (120-1).

13. The method according to claim 12, wherein the initiation of the cleaning process of the second filtration unit (120-2) comprises a selection of the cleaning process from the following group: cleaning by backwashing, cleaning-in-place, cleaning by backwashing and cleaning-in-place, General cleaning and predetermined cleaning programs.

14. The method of claim 1 1, wherein adjusting the control parameter of the filtering process of the second filtration unit (120-2) is a reversal of a flow direction of the Unfiltrats in the at least one membrane filter module (401) of the second filtration unit (120-2).

15. The method according to any one of claims 8 to 14, wherein the adaptive control of the occupancy of the second filtration unit (120-2) takes place on the basis of a fuzzy logic and / or artificial neural networks.