TECHNICAL FIELD

The present invention relates to a filtering method, in particular a filtering method for a food fluid.

Filtration is a very widely-used single operation for separating solids from fluids. A solid phase dispersed in a fluid phase is separated. This is normally done by causing the fluid to be filtered to be passed through a porous filtering element, which can retain the solid phase and allow the fluid phase to flow out.

BACKGROUND ART

At present there exist various types of filtering elements, which can be described or classified on the basis of the filtration mechanism they actuate. In this sense, two main types of filtration mechanisms can be distinguished: exclusion, or surface filtration, when the solid phase is retained on the surface of the filter element due to the geometry and dimensions thereof, i.e. it cannot pass through the filter element;

depth, when retaining the solid phase is achieved by trapping, for example mechanical, hydro-dynamic, chemical or another method, internally of the filter element.

In the field of alimentary products, and more specifically in the field of drinks, surface filtration is the most widely used, but in some cases a type of filtration is used that can be assimilated to depth filtration.

In the case of surface filtering, the dispersed solids deposit exclusively or nearly exclusively on the surface of the filtering element which is located on the side directly exposed to the fluid to be filtered, conventionally called the dregs side, progressively forming on the surface a layer of solid particles, conventionally called the cake.

Surface filtration is the preferred method when the solid dispersed phase tends to form a solid panel that is rigid and permeable to the fluid phase (cake filtration). This is the case, for example, in the filtration of tartrate crystals in refrigerated wines. If the dispersed solid phase tends to form a solid compressible panel that is not very permeable to the fluid phase, as in the case of colloidal dispersions, surface filtration is used only if the solid contents is relatively low, typically less than 1% of the overall weight of the fluid phase.

In depth filtration, the dispersed solids are prevalently retained internally of the filter element, i.e. in the thickness of the filtering element which is interposed between the surface exposed to the fluid to be filtered and the surface exposed to the fluid already filtered.

In the field of depth filtration four main retaining mechanisms are recognised: by direct intercepting and inertia impact, by diffusion, by interaction (for example electrostatic interaction), and by gravimetry. In general terms, given equal fluid-dynamic conditions, in depth filtration the largest particles are normally retained with inertia impact or intercepting mechanisms, while the smallest particles are retained by diffusion or electrostatic interaction. The intermediate-sized dimension particles are retained with a combination of the above-cited mechanisms.

In the industrial field, the filtration is performed by a plant which generally comprises a pump suitable for removing the fluid to be filtered from a tank and supplying it internally of a filtering device, which comprises a casing and at least a filtering element suitable for sub-dividing the internal volume of the casing into a first chamber communicating with an inlet of the fluid to be filtered and a second chamber communicating with an outlet of the filtered fluid, which is in turn connected to a collecting tank of the fluid.

This type of filtration can generally be performed following two different operating modes: constant differential pressure or constant flow rate.

In differential pressure filtering, the difference between the fluid pressure upstream of the filter element and the pressure of the filtered fluid downstream of the filter element remains constant during the filtration.

This operating mode can be performed by monitoring, via special pressure sensors, the pressure upstream and downstream of the filter element, and retroactively adjusting the pump, such that the difference between the monitored pressure values is at a set value. In constant-flow filtration the flow rate of the fluid crossing the filter element is constant during filtration. If the pump is volumetric, this operating mode can be achieved by simply setting a constant velocity on the pump. Otherwise it can be performed by monitoring, via a suitable flow rate sensor, the fluid flow rate which crosses the filter element, and retro-actively adjusting the pump, such that the value of the monitored flow-rate is at a set rate.

In relation to the above explanation, as the filtration progresses, the filter element progressively accumulates the solid phase dispersed in the fluid to be filtered, up to reaching a degree of clogging which does not enable filtration to continue, or because the fluid flow rate in outlet is too low and the filter times too long, or because the differential pressure becomes so high as to cause damage to the filter element.

In these cases, the filtration is halted and the filter plant is generally subjected to a washing step which includes activating a fluid (possibly a part of the previously-filtered fluid) to cross the filter element in counter-current, such as to detach and remove the accumulated solid phase, which is then eliminated through the opening of suitable discharge conduits.

One of the problems encountered when a filtration is performed, whether in constant differential pressure or in constant-flow, is the optimisation of the process parameters, i.e. the choosing of the differential pressure value, or respectively the flow rate value to be set in order to perform efficient filtering. These values generally depend on multiple factors, among which in particular the chemical-physical characteristics of the filter element and the fluid- dynamic characteristics of the fluid to be filtered.

This problem is conventionally solved by use of empirical tests on a small- scale, for example by using a small-scale model of the filter plant, with which values of the process parameters can be determined which will thereafter be set and used in the filtration operations on an industrial scale.

These tests, which in any case require a large investment of economic resources and time, often lead to erroneous evaluations and dimensions, or in any case not optimal evaluations, as in general they are performed on sample aliquots which can be unrepresentative of the behaviour of the real quantity of fluid which has to be filtered on an industrial scale.

DISCLOSURE OF INVENTION

In the light of these considerations, an aim of the present invention is to provide a filtration method which enables setting and modulating the values of the filter parameters, on the basis of the data recorded during filtration itself, such that the values are more realistic than those obtainable with the empirical tests as outlined above.

A further aim of the present invention is to attain the above-mentioned objective with a solution that is simple, rational and relatively economical. These aims are attained by the characteristics of the invention as reported in independent claim 1. The dependent claims delineate preferred and/or particularly advantageous aspects of the invention.

The invention is underpinned by laws of physics which described motions of fluids through a porous means, such as for example a filter element.

One of these laws is known as Darcy's law, which establishes a relation between the filtration flow rate and the resistance offered by the filter element, the filtrate viscosity, the filtration surface and the pressure difference detected upstream and downstream of the filter. A possible formulation of Darcy's law is expressed by the following equation:

dV _ A ^■ Ap

dt η - R

where dV/dt is the fluid flow rate which crosses the filter element, V is the fluid volume which crosses the filter element, t is the time, η is the viscosity of the fluid which crosses the filter element, A is the filtration surface of the filter element, R is the specific resistance of the filter element, and Δρ is the differential pressure applied to the filter element, i.e. the difference between the fluid pressure upstream of the filter element and the fluid pressure downstream of the filter element.

As this is a filtration, the preceding equation can be modified such as to take account of the resistance Rd offered by the progressive increase in the quantity of solids which deposit on the filter element according to the following equation: dV _ A - Ap

dt ^~ η - iR + Rd) or:

dt _ η - R η - Rd

dV ^~ A ^■ Ap A - Ap ^'

In the usual case in which the filter element is destined to perform a surface filtration, and in the hypothesis that all the solids of the fluid to be filtered are retained by the filter element, the term Rd can also be expressed by the relation:

Rd = V ^■ c ^■ a

and thus the preceding relation becomes:

where a is the specific resistance offered by the solids deposited on the filter element expressed per unit of solid mass, and c is the concentration of solids in the fluid to be filtered.

In the case that the filtration is done at a constant differential pressure, the term Δρ is a constant, such that the equation [1] can be approximated to the equation of a straight line:

- = cl - V + c2 [2]

V

where d and c2 are two constants that depend on the filtration surface, the viscosity and the solid content dispersed in the fluid, the specific resistance and the differential pressure applied on the filter means: η - c - a

c\ = e

A - Ap c2 = *- .

A - Ap

Should the filtration be done at a constant fluid flow rate, the term dV/dt = V/t = Q is a constant, and the equation [1] can be rewritten in one of the following forms:

Ap = k\ - t + k2 [3] or

Ap = k3 - V + k4 [4] where k1 , k2, k3, and k4 are constants: &1 = ^J Q

A

k3 = ^^ . Q

A

A

Equations [3] and [4] are generally valid on the condition that the solids deposited on the filter element are non-compressible, such that the increase in resistance to the filtration is linearly proportional to the quantity of solids deposited on the surface of the filter element. In the majority of cases, however, the increase in the differential pressure between upstream and downstream of the filter element also leads to the compression of the deposited solids, an effect which in turn produces a further increase in resistance to filtration. Differently to the constant pressure of filtration, therefore, the progressive increase in resistance to filtration does not depend only on the quantity of solids deposited but also on the absolute pressure they are subject to. The increase in resistance to filtration due to the compression of the solid cake can be expressed by the following relation: a = ₀ ^■ (Ap) ⁿ , where the exponent n is a constant which depends on the characteristics of the solid retained and accumulated on the filter which, usually, is determined by empirical filtration results.

For relatively contained pressure differential variations, we can refer to a mean resistance value, which is calculated by integration, and expressed as: a = a ₀ - (l - n) - Δρ"

replacing the relation in equation [1], considering that the term ao is a constant, in the sense that it does not depend on the pressure but on the characteristics of the solids dispersed and retained by the filter means, grouping the constants and passing to the logarithms, equations [3] and [4] can be respectively rewritten in the following form: ln( = k5 ^■ 1 ^η (Δρ) - k6 [5] and

ln(V) = k7 - ln(Ap) - k$ [6] where k5, k6, k7, and k8 are constants:

k5 = l - n η - c

k6 = In (ι -«)·δ ² kl = \ - n

£8 = In

A

On the basis of the preceding considerations, the invention discloses a filtration method of a fluid, mainly a food fluid, which comprises the step of activating the fluid to cross at least a filter element such as to perform a filtration and, during the filtration, to perform a control cycle comprising following steps of:

monitoring the value of a first and a second process parameter of filtration chosen from among: time passed from the start of the control cycle, quantity of fluid which has passed though the filter element from the start of the control cycle, and pressure differential at the heads of the filter element, i.e. the difference between the pressure of the fluid upstream of the filter element and the fluid pressure downstream of the filter element,

using the monitored values of the first and second process parameters in order to determine the coefficients of a correlation function, preferably continuous, which sets in a single relation the first and second process parameter, for example a function of correlation based on Darcy's law as previously described,

selecting one from the first and second process parameters as a reference parameter.and the other as a control parameter,

setting a desired value of the reference parameter, for example a value of the reference parameter which it is desired to reach at the end of the filtration, applying the desired value and the previously-determined coefficients to the correlation function of the desired value, in order to estimate a corresponding value of the control value, which therefore represents the value of the control parameter which is expected at the end of filtration.

During this control cycle, the activation of the fluid is controlled such that an operating parameter of the filtration is constantly equal to or almost a predetermined value, the operating parameter being selected from between: the pressure differential at the heads of the filter element and the fluid flow rate crossing the filter element (13).

In other words, this control cycle can be performed in both filtration modes: at constant pressure differential, or at constant flow-rate.

The first mode comprises controlling the activation of the fluid to be filtered such that the differential pressure at the heads of the filter element, during the control cycle, is constantly the same (or almost) as a predetermined value.

The second mode comprises control of the activation of the fluid to be filtered such that the fluid flow rate crossing the filter element, during the control cycle, is constantly the same (or almost) as a predetermined value.

With a constant pressure differential filtration, the continuous function used in the control cycle can be expressed for example by the preceding equation [2]. In this case, the first process parameter can be the time t that has passed from the start of the control cycle, the second process parameter can be the quantity V (expressed in volume) of fluid which has crossed the filter element from the start of the control cycle. The coefficients to be determined can be c1 and c2.

The coefficients can be determined simply by means of steps of:

detecting, during the filtration of the control cycle, at least two pairs of corresponding values (t, V) of the first and second process parameters, applying the pairs of values corresponding to the correlation function [2], such as to obtain two equations having as unknowns the above-mentioned coefficients, and then

putting the equations in a system such as to determine the values of the coefficients.

A preferred aspect of the invention comprises:

detecting more than two copies of values of the first and second process parameter, preferably at least four or five pairs,

applying the pairs of corresponding values to the correlation function, such as to obtain a same number of equations having the coefficients as unknowns, using the equations in pairs in order to determine a plurality of values indicating the coefficients, and

calculating each coefficient as a mean of the relative indicative values.

After these coefficients of the correlation function [2] have been determined, the quantity of fluid can be selected as reference parameter and the time as control parameter, or vice versa. In the first case, it is thus possible to set a desired value of the quantity of fluid which must have ben filtered at the end of the filtration, and obtain from the function [2] an expected value of the duration of the filtration in order to reach the desired quantity of filtered fluid. In the second case, it is on the other hand possible to set a desired value of the duration of the filtration and obtain from the function [2] an expected value of the total quantity of filtered fluid at the end of the filtration.

Still with reference to a constant pressure differential filtration, an aspect of the invention comprises:

repeating the above control cycle various times, each time using a different predetermined value of the pressure differential,

selecting as an optimal value, from among the pressure differential values used during the various control cycles, the value for which the expected value of the control parameter is indicative of a higher mean flow rate, and proceeding with the filtration, controlling the activation of the fluid such as to maintain a constant pressure differential that is substantially equal to the selected optimal value.

After having determined the optimal value of the differential pressure, another aspect of the invention includes the possibility to use the coefficients of the correlation function, determined during the control cycle performed with the optimal value of the pressure differential, in order to determine an optimal number of washing steps to be performed during the filtration, and thus to proceed with the filtration, performing a number of intermediate washing steps which is equal to the optimal number thus determined.

In particular, the determination of the optimal number of washing steps can be performed by steps of:

performing a calculating cycle which comprises:

setting a hypothetical number of washing steps, preferably starting from zero, calculating a partial value of the quantity (typically expressed in volume) of fluid to be filtered on the basis of a total value of the quantity of fluid which has to be filtered at the end of the filtration and the hypothetical number of washing steps, for example by dividing the total volume by a hypothetic number of washes increased by one, i.e. by the number of partial filtration steps which it can be hypothesised will be performed between the washing steps,

using the correlation function, the relative coefficients determined by the optimal value of the pressure differential and the partial value of the quantity of fluid to be filtered, in order to estimate a partial value of a filtration time, which will correspond to a predicted duration of each partial filtration step, calculating a total value of the filtration time on the basis of the partial value, the hypothetical number of washing steps and a predetermined value of the duration of each of the washing steps,

repeating the calculation cycle, increasing each time the hypothetical number of washing steps (preferably of one unit at a time), up to when the total filtration time determined with the latest cycle of calculation is greater than the time determined with the preceding calculation cycle of up to when the hypothetical number of washing steps is equal to a maximum admissible number, then

setting, as an optimal number of washing steps, the number set for each preceding calculating cycle, or respectively,

setting, as an optimal number of washing steps, the maximum admissible number.

In the context of a constant flow filtration, the continuous function used in the control cycle can be expressed for example by one of equations [3], [4], [5] or [6], previously described. In this case, the first process parameter can thus be the pressure differential Δρ, while the second process parameter can be selected from among: the time t that has passed since the start of the control cycle and the quantity V (expressed in volume) of alimentary fluid which has crossed the filter element from the start of the control cycle. The coefficients to be determined can be respectively k1 and k2, k3 and k4, k5 and k6, or k7 and k8.

In this case too, the coefficients can be determined simply via the steps of: detecting, during filtration in the course of the control cycle, at least two pairs of corresponding values, for example (Δρ, V) of the first and the second process parameter, applying the pairs of values corresponding to the function of correlation, such as to obtain two equations having as unknowns the above-mentioned coefficients, and then

putting the equations in a system to determine the values of the coefficient. A preferred aspect of the invention however comprises, in this case too:

detecting more than two pairs of corresponding values of the first and second process parameters, preferably at least four or five pairs,

applying the pairs of values corresponding to the function of correlating, such as to obtain a same number of equation having as unknowns the above- mentioned coefficients,

placing the equations in pairs in a system to determine a plurality of indicative values of the coefficients, and

calculating each coefficient as a mean of the relative indicative values.

After the coefficients have been determined, the differential flow rate can be selected as a reference parameter and the time or quantity of fluid as control parameter. In this way a maximum desired value of the pressure differential to be reached can be set at the end of the filtration, and via the function of correlating it is possible to obtain an expected value of the duration of the filtration or an expected value of the total quantity of fluid filtered, before reaching the desired pressure differential.

In this case too, an aspect of the invention comprises:

repeating the above control cycle several times, using a different predetermined value of the flow rate each time,

selecting as optimal value, from among the flow rate values used during the various control cycles, the value for which the expected value of the control parameter is indicative of a higher total quantity of filtered fluid,

proceeding with the filtration, controlling the activation of the fluid such as to maintain a constant flow rate that is substantially equal to the optimal selected value.

After having determined the optimal value of the flow rate, in this case too it is possible to use the coefficients of the function of correlation determined during the control cycle, performed with the optimal flow rate, in order to determine an optimal number of washing steps to be performed during the filtration, and thus proceeding with the filtration by performing a number of intermediate washing steps equal to the optimal number thus determined. In particular, the determination of the optimal number of washing steps is preferably performed by means of the steps of:

performing a cycle of calculation which comprises:

setting a hypothetical number of washing steps, preferably starting from zero, calculating a partial value of the quantity (typically expressed in volume) of fluid to be filtered on the basis of a total value of the fluid quantity which must have been filtered at the end of the filtration and the hypothetical number of washes, for example dividing the total volume by the hypothetical number of washes increased by one, i.e. by the number of partial filtration steps which it is hypothesised to perform between the washing steps,

using the correlation function, the coefficients determined for the optimal value of the flow rate and the partial value of the quantity of fluid to be filtered, in order to estimate a corresponding partial value of a filtration time, which corresponds to an anticipated duration of each partial filtration step, calculating a total value of the filtration time on the basis of the partial value, the hypothetical number of washing steps and a predetermined value of the duration of each of the washing steps,

repeating the calculating cycle, increasing each time the hypothetical number of washing steps (preferably by one unit at a time), up to when the total filtration time, determined with the last calculating cycle, is equal to or greater than the one determined with the preceding calculation cycle or up to when the hypothetical number of washing steps is equal to a maximum admissible number, and then:

setting, as the optimal number of washing steps, the step set for the preceding calculating cycle, or respectively,

setting as the optimal number of washing steps the maximum admissible number.

It is specified that the equations [2], [3], [4], [5] and [6] might be specially modified in order to take into account other factors which might intervene during filtration. They would however be scientific or mathematical modifications and therefore have no influence on the technical aspects of the control cycle, and are therefore considered to be comprised in the present invention. In the same way, as far as the present invention is concerned, the function of correlation used in the control cycle might not be based on Darcy's law, but on another law which describes the physical phenomenon of a fluid crossing a porous element.

The filtration method of the present invention can be performed with the aid of a computer program comprising an information code for performing all the steps of the method described herein above, and can possibly be actuated in the form of an IT product comprising the above-mentioned computer program.

The present invention discloses a filter apparatus for a fluid, typically an alimentary fluid, comprising a casing having an inlet and an outlet, at least a filter element located such as to subdivide the internal volume of the casing in a first chamber communicating with the inlet and a second chamber communicating with the outlet, first measuring means for monitoring a time elapsed, second measuring means for monitoring a quantity of fluid which crosses the filter element, third measuring means for monitoring a difference between the fluid pressure at the inlet and the fluid pressure at the outlet, and a control device configured such as to perform a method according to any one of the preceding claims.

For example, the control system can comprise a processing unit, for example a microprocessor, a data storage unit connected to the processing unit, and the computer program mentioned herein above, which is memorised in the data storage unit, such that when the processing unit performs the computer program, all the steps of the filtration method are performed.

BRIEF DESCRIPTION OF DRAWINGS

Further characteristics and advantages of the invention will emerge from a reading of the following description provided by way of non-limiting example, with the aid of the figures illustrated in the accompanying tables of drawings. Figure 1 is a diagram of a filter apparatus according to the present invention. BEST MODE FOR CARRYING OUT THE INVENTION

Figure 1 schematically illustrates a filter apparatus 10 for a food fluid, typically wines or other enological fluids of various nature.

The filter apparatus 10 mainly comprises a filter device or a filter 11 , which in particular comprises a casing 12 containing a plurality of elements of filter cartridges 13. The casing 12 is associated to an inlet conduit 14 of the fluid to be filtered and an outlet conduit 15 of the filtered fluid. The inlet conduit 14 is designed to be connected to a storage tank 16 of the fluid to be filtered, while the outlet conduit 15 is destined to be connected to a collection tank 17 of the filtered fluid.

In the illustrated example, the casing 12 is formed by an upper bell 120 closed on the bottom by a lower base 121. The filter elements 13 are singly conformed as tubular bodies, internally hollow, located vertically and having a closed upper end and an open lower end. The lower end of each filter element 13 is rested on and closed by the lower base 121 , which in turn comprises a plurality of openings (not illustrated) which are singly destined to set the internal cavity of a relative filter element 13 in communication with the outlet conduit 15.

In this way, the filter elements 3 subdivide the internal volume of the casing 12 into two distinct chambers, of which a first chamber 18, defined between the external surface of the filter elements 13 and the internal surface of the bell 120, which is set in direct communication with the inlet conduit 14, and a second chamber 19, defined by the internal cavities of the filter elements 13, which is set in direct communication with the outlet conduit 15.

The filter elements 13 can be of various types and materials, for example cartridges made of polymer, ceramic, sintered materials, lenticular materials, etc.

A supply pump 20 is located along the inlet conduit 14, which is for pumping the fluid to be filtered from the storage tank 16 towards the filter device 1 . In this way, the fluid to be filtered is normally forced to cross the filter elements 13 of the filter device 11 , which retain the solid phase dispersed therein and allow the liquid phase to pass, i.e. by subjecting the fluid to a filtering operation. The supply pump 20 can be a self-priming pump and operating at a low number of revolutions, typically comprised between 200 and 900 revolutions per minute.

Between the supply pump 20 and the filter device 11 , the inlet conduit 14 comprises a prefilter 21 , the function of which is to operate a first rough filtration of the fluid to be filtered, such as to retain the largest solid particles which otherwise might damage or clog the filter elements 13 too rapidly. The pre-filter 21 can be a bag filter with a filtration degree comprised between 20 and 200 micron.

The filtration apparatus 10 comprises at least two pressure transducers, of which a first pressure transducer 22 positioned on the inlet conduit 14 downstream of the prefilter 21 , and a second pressure transducer 23 positioned on the outlet conduit 15. These pressure transducers 22 and 23 are destined to respectively measure the pressure of the fluid to be filtered upstream of the filter device 11 and the pressure of the filtered fluid downstream thereof. The pressure transducers 22 and 23 are connected to a process logic control (PLC) 60, which schematically comprises at least a processing unit 600, typically a microchip, and a data storage unit 601 connected with the processing unit 600. The PLC 60 will use the measurements supplied by the pressure transducers 22 and 23 to calculate a value of the pressure differential at the heads of the filter device 1 1 , i.e. a value of the pressure difference between upstream and downstream of the filter elements 13. For a more accurate measurement of the pressure upstream and downstream of the filter device 1 1 , and in particular for a more accurate calculation of the relative pressure difference, the pressure transducers 22 and 23 must preferably be set at a same height from the ground. If this is not so, the PLC 60 will have to take account of the pressure exerted by the fluid column extending along the difference in level between the two pressure transducers 22 and 23. The transducers 22 and 23 must preferably have a minimum sensitivity of 0.01 bar.

The PLC 60 is further connected to a flow rate measurement device 24, which is located on the outlet conduit 15 such as to measure the flow rate of filtered fluid which flows towards the collecting tank 17. The measuring device 24 is provided with an analog and pulse transmitter which enables the PLC 60 to detect both an instant flow rate of filtered fluid and a cumulative value of the quantity (typically expressed in volume) of fluid which has been filtered starting from a predetermined initial instant.

Thanks to the aid of the above-mentioned pressure transducers 22 and 23 and the flow rate measuring device 24, the PLC 60 is programmed such as to control the functioning of the supply pump 20 such that the filtration can be done at a constant differential pressure or in a constant flow rate mode.

The first mode can comprise that during filtration the PLC 60 monitors the value of the pressure differential at the heads of the filter device 1 1 , by means of the pressure transducers 22 and 23; that it calculates the difference or error between the monitored value and a predetermined value required by the differential pressure; and that is retroactively regulate the velocity of the supply pump 20 such as to minimise the error.

The second mode can similarly comprise that during filtration the PLC 60 monitors the instant value of the filtered fluid flow rate downstream of the filter device 1 1 , by means of the flow rate measurement device 24; that it calculates the difference or error between the monitored value and a predetermined value required by the flow rate; and that it retroactively regulates the velocity of the supply pump 20, such as to minimise the error. In a case in which the supply pump 20 is a volumetric pump, such that the velocity thereof is strictly proportional to the flow rate delivered, the second mode might be carried out with an open-chain control cycle, i.e. the PLC 60 might simply activate the supply pump 20 at the velocity corresponding to the required flow rate value.

In the illustrated example, the filter apparatus 10 comprises a washing system of the filter elements 13, which comprises a tank 25 into which four inlet conduits 26 flow which are provided with respective check valves 27, of which a first conduit destined to be connected to a cold water source, a second conduit destined to be connected with a hot water source, and two further conduits destined to be connected, via repsective pumps 28, to a like number of tanks 29 of a special detergent. The check valves 27 and the pumps 28 are commanded by the PLC 60, which is programmed to form, internally of the tank 25, a washing mixture. The tank 25 is provided with an outlet conduit 30 which is provided with a check valve 31 and is connected to the inlet conduit 14 upstream of the supply pump 20. The washing system further comrpises a first bypass conduit 32 suitable for connecting the inlet conduit 14 downsteram of the prefilter 21 with the outlet conduit 15, and three check valves 33, 34 and 35, respectively positioned on the first bypass conduit 32, on the inlet conduit 14 upstream of the entry point of the first bypass conduit 32, and on the outlet conduit 15 upstream of the entry point of the first bypass conduit 32. The washing system also comprises a second bypass conduit 36 which connects a point of the inlet conduit 14 upstream of the valve 34 with a point of the outlet conduit 15 upstream of the valve 35, and a check valve 37 positioned along the second bypass conduit 36.

To carry out the washing of the filter elements 13, the PLC 60 closes the communication between the inlet conduit 14 and the storage tank 16 of the fluid to be filtered, maintains the supply pump 20 in function up until the outlet conduit 15 of the filtered fluid is emptied, and finally sets the outlet conduit 15 in communication with a discharge. At this point, the PLC 60 opens the valve 31 and newly activates the supply pump 20, such as to supply the washing fluid towards the filter device 1 1. By suitably commanding the check valves 33-35 and 37, the PLC 60 can cause the washing fluid to flow across the filter elements 13 in the same current direction with respect to the fluid to be filtered, i.e. in the same direction in which the filtration occurs (valves 34 and 35 open; valves 33 and 37 closed), or in counter-current with respect to the fluid to be filtered (valves 34 and 35 closed; valves 33 and 37 open). In the latter case, the washing fluid originating from the supply pump 20 first runs through the second bypass conduit 36, re-enters the filter device 1 1 in counter-current, passes in the first bypass conduit and finally returns into the outlet conduit 15. The solid cake which forms on the surface on the dregs side of the filter elements 13 is then detached and discharged out of the filter device 1 1 through the outlet conduit 15 towards the discharge. When the washing step has been completed, the PLC 60 returns the filter apparatus 10 into the configuration in which it is destined to filter the fluid contained in the storage tank 16, as previously described.

In this context, when the above-described filtration apparatus 10 is used to perform a filtration, the PLC 60 is programmed such as to perform the control method of the filtration described in the following.

This control method first comprises that the PLC 60 must know whether the filtration has to be performed in constant pressure differential mode or in constant flow-rate mode. This information can bne provided directly by an operator, for example by means of an interface system (such as a keyboard, a touchscreen monitor or similar) connected to the PLC 60.

In the case in which the filtration has to be performed in a constant pressure differential, the control method comprises that the PLC 60 also acquires:

- a total value VJot of the volume of fluid which has to be filterd by the end of filtration,

- a minimum admissible value Q_min of the fluid flow rate that crosses the filter device 11 ,

- a minimum admissible value Ap_min of the pressure differential at the heads of the filter device 1 ,

- a maximum admissible value Ap_max of the pressure differential at the heads of the filter device 11 ,

- a maximum admissible number N_max of the number of washing steps which can be carried out during the filtration,

- an expected value tjav of the duration of each of the washing steps.

These values too can be provided directly by the operator, for example via the above-mentioned interface system connected to the PLC 60.

At this point, the PLC 60 starts up the supply pump 20 such as to begin the filtration. During this filtration, the PLC60 is able to continuously monitor:

the value of the pressure differential at the heads of the filter device 1 , the value of the fluid flow-rate which crosses the filter device ,

the value of the cumulative volume of fluid which has been filtered.

The value of the pressure differential can be calculated by means of the pressure transducers 22 and 23, while the values of the intant flow rate and the cumulative volume can be determined by means of the flow rate measuring device 24.

Immediately after the start of the filtration, the PLC 60 performs a filtration control cycle which firstly controls the functioning (the velocity) of the supply pump 20 so that the filtration, during the whole control cycle, is conducted at a predetermined constant pressure differential, the value of which can be denoted generically as Δρ.

Starting from the beginning of the control cycle, the PLC 60 continuously monitors:

the time t that has passed from the start of the control cycle, for example using a timer, and

the value V of the cumulative volume of fluid which has been filtered from the start of the control cycle.

After having waited a time necessary for stabilising the system (which can be set by the operator), the PLC 60 acquires at least four or five pairs of values of the parameters mentioned above, generically denoted as (t _j , V _j ), relative to successive instants j during the course of the control cycle. In other words, the term t _j is the time that, at any generic instant j, has passed since the start of the control cycle, while the term V _j is the cumulative fluid volume which, at the instant j has been filterd from the start of the control cycle. In order to obtain each pair of values, the PLC 60 can acquire the time t _j required for filtering with a constant pressure differential of Δρ, a prefixed volume V _j , of fluid, or, vice versa, it can acquire the volume V _j of fluid that has been filtered after a predefined time t, after the start of the control cycle.

Each pair of values (t _j , V _j ) is then applied to the equation [2] described above, obtaining:

— = cl - V, + c2 in which the only unknown quantities are the coefficients c1 and c2.

The equations obtained for each pair of values (t _j , V _j ) are used in pairs, such as to calculate n pairs of values (c1 _m , c2 _m ) of the unknown coefficients. By means of these pairs of values, the PLC 60 calculates the definitive values of the unknown coefficients of the equation [2] via the following means:

∑ l _m

c ^

C2 =

At this point, the PLC 60 applies, to the equation [2], coefficients c1 and c2 obtained with the preceding means, and the value V_tot of the fluid volume that will have been filtered at the end of the filtration, such as to calculate an expected value t_prev of the total duration that the filtration would have should a constant differential pressure of Δρ be maintained, equal to the one used in the control cycle.

As it knows the expected value t_prev of the total duration of the filtration, the PLC 60 also calculates an expected value Q_prev of the mean flow rate of the filtered fluid, with the following equation:

^_ t __ prev

Finally, the PLC 60 uses the coefficients c1 and c2 calculated with the preceding means, and the value Q_min of the minimum flow rate, in order to calculate a value V_max of the cumulative volume of filtered fluid, using the following equation: — = cl - F_max+ c2 .

g rnin

This equation descends directly from the equation [2], and the value V_max thus obtained represents the maximum volume of fluid that can be filtered with a constant pressure differential of Δρ and with a mean flow rate of the minimum admitted flow rate Q min. The value Q_min can be expressed both in absolute value as a percentage x% of the value Qo of the fluid flow rate initially detected for the pressure differential Δρ considered, which is linked to the absolute value of the following relation:

This control cycle is repeated at least twice starting from the start of filtration. For the first control cycle, the value Δρ of the pressure differential is preferably set equal to the minimum value Ap_min previously acquired: Δρ= Δρ_ιτιίη. For each following control cycle, the value Δρ of the pressure differential is set such as to be greater than the one set in the preceding cycle, for example increased by a predetermined quantity.

Any generic control cycle (/-th) will therefore be performed at a pressure differential Δρ, great than the Δρ _Μ of the preceding control cycle (Δρ _ί >Δρ _ί-1 ), and will provide values t_prev, of the envisaged filtration duration, Q_prev, of the envisaged mean and a V maxj of the maximum filtrable volume, corresponding to the relative value Δρ, of the pressure differential.

After each control cycle, the PLC 60 compares the value of t_prev, of the filtration duration with the t_previ.i calculated in the preceding control cycle. If t_previ is less than tjDreVj.-ι, the PLC 60 carries out a further control cycle, and regulates the functioning (velocity) of the supply pump 20 in such a way as to proceed with the filtration with a pressure differential value Δρ, ₊ ι that is greater with respect to the value Δρ,; or otherwise it interrupts execution of the control cycle, selects the value Δρ _Μ used in the preceding control cycle as optimal value Δρ* of the differential pressure, and proceeds with the filtration, controlling the functioning (velocity) of the supply pump 20 such as to maintain a constant differential pressure which is equal to the optimal value Δρ*=Δρ _Μ selected.

Alternatively, after each control cycle, the PLC 60 could compare the value Q_prev, of the mean fluid flow rate with the Q_prevj.-i calculated in the preceding control cycle. If Q_prev, is greater than Q_prev _i-1 , the PLC 60 carries out a further control cycle, regulating the functioning (velocity) of the supply pump 20 such as to proceed with the filtration with a differential pressure value Δρ,+ι that is greater than value Δρ,; otherwise it interrupts the control cycle, selects the value Δρ used in the preceding control cycle as optimal value Δρ ^* of the pressure differential, and proceeds with the filtration, controlling the functioning (velocity) of the supply pump 20, such as to maintain a constant pressure differential that is constant and substantially equal to the optimal value Δρ ^* =Δρ _ί-1 selected.

Once the optimal value Δρ ^* =Δρ _Μ of the pressure differential has been established, the PLC 60, in the control method, establishes whether it will be necessary during the course of the filtration to perform one or more washing steps of the filter elements 13. This decision is made by the PLC 60 by means of the comparison of the total value V_tot of the quantity of fluid which will have to be filtered at the end of the filtration with the value V_maxj.i of the maximum filtrable volume calculated during the control cycle carried out at the optimal pressure differential Δρ _Μ =Δρ ^* . In particular, the PLC 60 establishes if one or more washing steps are required, if the value V_maXj.i is below the value V_tot.

If the PLC 60 establishes that one or more intermediate washing steps are required, the number of these steps is established and optimised with the procedure described in the following.

In the procedure, the PLC 60 performs a calculation cycle comprising the initial step of setting a hypothetical number N of washing steps.

In the hypothetical number N of intermediate washing steps, hypothetically the filtration will be overall subdivided into a number of partial filtration steps which is equal to the hypothetical number N of intermediate washes, increased by one unit. Consequently, the volume of product that will be filtered during the course of each partial filtration step will be equal to the total volume of fluid which remains to be filtered, divided by the hypothetical number of partial filtration steps.

The calculating cycle thus comprises calculating a partial value V_fase of the quantity (typically expressed in volume) of fluid to be filtered at each partial filtration step, according to the following relation:

N + l

in which V_tot ^* is equal to the total value V_tot of fluid that has to be filtered at the end of filtration, from which the quantity of fluid already filtered since the start of filtration has been subtracted, in particular that which has been cumulatively filtered during the preceding control cycle.

The value V_fase thus obtained is used to calculate a predicted value t_prev_fase according to the following equation: t _ prev _ fase = cl - V _ fase ² + c2 · V _ fase which derives from the equation [2] described herein above, and in which the coefficients C1 and c2 are those calculated (using the means) during the control cycle performed at the optimal pressure differential Δρ*=Δρ _ί-1 .

At this point, the calculation cycle comprises the PLC 60 calculating an expected value t_prev_tot of the overall duration of the filtration, according to the following equation:

t _ prev _tot = (N + 1) · t _ prev _ fase + N - t _lav in which tjav is the duration of each washing step of the filter elements 13, as set by the operator.

The above-described calculation cyle is repeated at least twice by the PLC 60. For the first calculation cycle, the hypothetical number N of washing phases is preferably set at zero: N=0. For each successive calculation cycle, the hypothetical number N of washing phases is increase by a unit with respect to the preceding calculation cycle.

Each generic calculation cycle (K-th) will therefore by characterised by a hypothetical number N _K of washing phases greater than the number Νκ-ι of the preceding cycle (Νκ=Νκ-ι+1), and will give a value t_prev_tot _k of the overall filtration time corresponding to the relative hypothetical number N _K of the washing phases.

After each calculation cycle, the number N _K of washing cycles is compared with the maximum admissible number of washing cycles N_max set by the operator, while the expected value t_prev_tot _k of the overall filtration time is compared with the value t_prev_tot _k -i calculated with the preceding calculation cycle.

If t_prev_tot _k is greater than or equal to t_prev_tot _k- i, or if N _k is equal to N_max, the PLC 60 stops performing further calcuation cycles: otherwise, it performs a new calculation cycle, increasing the hypothetical number of washing cycles of a unit, as described herein above.

In particular, if t_prev_tot _k is greater than or equal to t_prev_tot _k- i, the PLC 60 sets the number N _k-1 of the preceding cycle as optimal number N ^* of washing phases: N ^* = N _k-1 . If on the other hand N _k is equal to Njnax, the PLC 60 sets the number N _k of the present cycle as optimal number N ^* of washing phases: N*= N _k- i.

Once the optimal number N* of intermediate washing steps has been established, the PLC 60 commands the filtration apparatus 10 to continue filtering with the optimal value Δρ* of the pressure diffemtial and to effectively carry out a number of intermediate washing steps that is equal to the optimal number established N ^* . ,

In the alternative case in which the filtration is to be performed in constant- flow mode, the control method comprises the PLC 60 acquiring:

a total value V_tot of the fluid volume which has to be filtered at the end of the filtration,

a minimum admissible value Apjnin of the pressure differential at the heads of the filter device 11 ,

a maximum admissible value Ap_max of the pressure differential at the heads of the filter device 11 ,

an admissible value Q_min of fluid flow rate which crosses the filter device 11 ,

a maximum admissible number N_max of the number of washing steps which can be carried out during filtration,

an envisaged value tjav of the duration of each of the washing steps.

These values can be directly supplied to the operator, for example by means of the interface system connected to the PLC 60.

In this case too, the PLC 60 starts up the supply pump 20 such as to start filtration. During this filtration, the PLC 60 is able to continuously monitor: the present value of the pressure differential at the heads of the filter device 1 1 ,

the present value of the fluid flow rate crossing the filter device 1 ,

the present value of the cumulative fluid volume which has been filtered. The present value of the pressure differential can be calculated by means of the pressure transducers 22 and 23, while the present values of the instant flow rate and the cumulative value can be determined by means of the flow rate measuring device 24.

Immediately after the start of the filtration, the PLC 60 performs a control cycle of the filtration which primarily comprises controlling the functioning (the velocity) of the supply pump 20 such that the filtration, during the whole control cycle, is performed at a predetermined constant fluid flow which crosses the filter device 1 1 , the value of which can generically be indicated as Q.

Starting from the beginning of the control cycle the PLC 60 continuously monitors:

the value V of the cumulative fluid volume which has been filtered since the start of the control cycle, and

the value Δρ of the pressure differential at the heads of the filter device.

After waiting for a time necessary for stabilising the system (settable by the operator), the PLC 60 acquires at least four or five pairs of values for the above-mentioned parameters, generically indicated as (V _j , Ap _j ), relative to successive instants j during the course of the control cycle. In other words, the term Vj is the cumulative fluid volume which, at generic instant j , has been filtered since the start of the control cycle, while the term Ap _j is the pressure differential which, at the same instant j, is calculated at the heads of the filter device 1. In order to obtain each pair of values, the PLC 60 can acquire the value V _j of the fluid volume that has been filtered on reaching a pressure differential of Ap _j or, vice versa, it can acquire the value Δρ _| of the pressure differential reached after a predetermined value Vj of the fluid has been filtered since the start of the control cycle.

Each pair of values (V _j , Apj) is applied to the equation [6] described in the preamble, obtaining: \n(V _J ) = k7 - ln(Ap _J ) - k8 in which the only unknowns are the coefficients k7 and k8.

The equations obtained for each pair of values (V _j , Ap _j ) are placed in the system in pairs, such as to calculate n pairs of values (k7 _m , k8 _m ) of the unknown coefficients. Using these pairs of values, the PLC 60 calculates the definitive values of the unknown coefficients of the equations [6] by means of the following means:

n

∑M _m

k8 = ^≤ .

n

Note that in place of the equation [6], the PLC 60 could use, in a similar way, equations [3], [4] or [5]. In the case of equations [3] and [4], the PLC 60 should naturally keep monitored, in the place of the values V _j of filtered fluid volume, the times t _j that have lapsed since the start of the control cycle. The choice of the equation to be considered has no influence on the control method as, since the flow rate Q is constant, the PLC 60 can always pass from the times filed to the filtered volumes field, or vice versa, simply by using the value Q of the flow rate, or respectively the inverse, as a conversion factor.

Returning to the example illustrated herein, the PLC 60 applies to the equation [6] the coefficients k7 and k8 obtained with the preceding means, and the value Apjnax of the pressure differential, such as to calculate an expected value V_prev of the cumulative quantity (expressed in volume) of fluid which would be filtered at the end of the filtration, such a constant flow equal to the flow rate Q used in the control cycle be maintained.

As the expected value V_prev of the total duration of the filtration is known, and as the flow rate Q is constant, the PLC 60 also calculates an expected value t_prev of the duration of the filtration in order to reach the predetermined value Ap_max of the pressure differential, with the following equation:

V prev

t prev =— = .

Q

This control cycle is repeated at least twice starting from the beginning of the filtration. For the first control cycle, the value Q of the fluid flow rate to be filtered is preferably set equal to the minimum value Q_min previously acquired: Q= Q_min.

For each successive control cycle, the value Q of the flow rate is set such as to be greater than the one set for the preceding cycle, for example increased by a predetermined quantity.

Any single generic control cycle (/-th) will therefore be carried out at a flow rate Q, that is greater than the QM of the preceding control cycle (Q^QM ) , and will provide values t_prev, of the envisaged filtration duration and V_prev, of the filtrable fluid volume corresponding to the relative value Q, of the flow rate.

After each control cycle, the PLC 60 compares the value of V_prev, of the volume of filtrable fluid with the value V_preV|.i calculated in the preceding control cycle. If V_prev, is greater than V_prevj.i , the PLC 60 performs a further control cycle, regulating the functioning (the velocity) of the supply pump 20 such as to proceed with the filtration with a flow rate value Q _j+ i that is greater with respect to the value Q,; otherwise, it interrupts the performing of the control cycle, selects the value Q used in the preceding control cycle as the optimal value Q* of the flow rate, and proceeds with the filtration, controlling the functioning (velocity) of the supply pump 20, such as to maintain a constant fluid flow rate which is substantially equal to an optimal selected value Q*=QM .

Once the optimal value Q ^* =Q _i-1 of the fluid flow rate has been established, the method comprises that the PLC 60 establishes whether it will be necessary, during the proceeding of the filtration, to perform one or more step of washing the filter elements 13. This decision is performed by the PLC 60 via the comparing of the total value V_tot of the quantity of fluid which will have to be filtered at the end of the filtration with the value of the filtrable volume calculated during the control cycle performed at the optimal flow rate Qi.i =Q*. In particular, the PLC 60 establishes where one or more than one washing steps are necessary, if the value V_preVj.i is lower than the value V_tot.

In a case in which the PLC 60 establishes that one or more intermediate washing steps are necessary, the number of these steps is established and optimised with the same procedure as the one described herein above.

This procedure comprises that the PLC 60 performs a calculation cycle comprising the initial step of setting a hypothetical number N of washing steps.

As previously explained, the hypothetical number N of intermediate washing steps means that hypothetical^ the filtration will be overall subdivided into a number of partial filtration steps which is equal to the hypothetical number N of intermediate washes increased by one unit. Consequently, the volume of product which will be filtered during the course of each partial filtration step will be equal to the total fluid volume which remains to be filtered, divided by the hypothetical number of partial filtration steps.

The calculation cycle comprises calculating a partial value V_fase of the quantity (typically expressed in volume) of fluid to be filtered in each partial filtration step, according to the following relation:

V rase =

N + l

in which V_tot* is equal to the total value V_tot of fluid which has to be filtered at the end of the filtration, from which has been substracted the quantity of fluid that has already been filtered since the start of the filtration, in particular the quantity which has been cumulatively filtered during the preceding control cycle.

The value VJase thus obtained is used to calculate an envisaged value t_prev_fase according to the equation:

- V fase

t _ prev _ jase =— =

Q *

in which Q ^* is the optimal value of the previously-established fluid flow rate to be filtered: Q ^* =Q _M .

At this point, the calculation cycle comprises that the PLC 60 calculates an expected value t_prev_tot of the overall duration of the filtration according to the following equation:

t _ prev tot = (N + 1) ^■ t _prev _ fase + N ^■ t _lav in which tjav is the duration of each washing step of the filtering element 13, as set by the operator.

The calculating cycle as described above is repeated at least twice by the PLC 60. For the first calculating cycle, the hypothetical number N of washing steps is preferably set at zero: N=0. For each successive calculating cycle, the hypothetical number N of washing steps is increased by a unit to the preceding calculating cycle.

Each generic calculating cycle (/c-th) will therefore be characterised by a hypothetical number N« of washing steps greater than the one N _K- i of the preceding cycle (Νκ=Ν _κ- ι+1), and will provide a value t_prev_tot _k of the overall filtration time, corresponding to the relative hypothetical number N _K of the washing steps.

After each calculating cycle, the number N _K of washing steps is compared with the maximum admissible number of washing cycles N_max set by the operator, while the expected value t_prev_tot _k of the overall filtration time is compared with the one t_prev_tot _k- i calculated with the preceding calculating cycle. If t_prev_tot _k is greater than t_prev_tot _k- i , or if N _k is the same as N_max, the PLC 60 stops performing further calculating cycles; otherwise, it performs a new calculating cycle, increasing the hypothetical number of washing cycle by a unit, as described herein above.

In particular, if t_prev_tot _k is greater than or equal to t_prev_tot _k- i , the PLC 60 sets the number N _k- i of the preceding cycle as the optimal number N ^* of washing steps: N ^* = N _k-1 . If on the other hand N _k is equal to N_max, the PLC 60 sets the number N _k of the present cycle as an optimal number N* of washing steps: N ^* = N _k- .

Once the optimal number N ^* of intermediate washing steps has been established, the PLC 60 commands the filtering apparatus 10 to continue the filtration with the optimal value Q ^* of the fluid flow rate to be filtered and to perform a number of intermediate washing stages which is equal to the established optimal number N ^* .

Obviously an expert in the field might make numerous modification of a technical-applicational nature to what is described herein above, without its forsaking the ambit of the invention as claimed herein below.