KR101269253B1 - Membrane cleaning interval prediction method and system in membrane bioreactor - Google Patents

Abstract

The present invention provides a linear model for predicting the intermembrane pressure difference of the biofilm reactor, the prediction time of the membrane module through the prediction of the intermembrane pressure difference, and provides a method and system for predicting the precise cleaning time of the membrane module. . In addition, we can build a predictive model using statistical least-squares method and cyclic least-squares method, and examine its effectiveness by applying RMSE to the predictive model.
In the method for predicting membrane cleaning timing of a biofilm reactor according to the present invention, in the method for predicting membrane cleaning timing of a biofilm reactor having an exponential relationship between the intermembrane pressure difference and time, the estimation index linearization of the intermembrane pressure difference and time curve is performed. 1 step of deriving a linear relationship; Deriving a measurement value by measuring a biofilm reactor operation time and an intermembrane pressure difference at a predetermined time period; Deriving a parameter of the linear relation whenever the measured value is derived by applying a cyclic least square method to the measured time and the interlayer pressure difference; And four steps of deriving the membrane cleaning timing by substituting the intermembrane pressure difference when the membrane cleaning is required in the linear relational expression.

Description

Method and system for predicting membrane cleaning time of biofilm reactor {MEMBRANE CLEANING INTERVAL PREDICTION METHOD AND SYSTEM IN MEMBRANE BIOREACTOR}

The present invention relates to a method and a system for predicting membrane cleaning time of a biofilm reactor, in particular, by linearizing the estimated indicators, linearizing the relationship between pressure differentials and deriving a parameter using a least square method or a cyclic least square method. A method and system for predicting the timing of membrane cleaning of a reactor.

The activated sludge process supplies air from an air supply tank (aeration tank) together with a bacterial community that oxidizes sewage under activated sludge conditions, and wastewater treatment in which colloidal or dissolved substances in sewage precipitate or are adsorbed on activated sludge. It is one of the construction methods. In this activated sludge process, the activated sludge which has been purged is returned to the main treatment process, and a membrane module is used to remove the secondary solids, which is called a membrane bioreactor (MBR).

The biofilm reactor is to supply oxygen from the bioreactor and stir the bacterial community to purge, and to pass the purified water through the membrane module to filter the sludge and solids so that the secondary purification.

Biofilm reactors can be categorized into sidestream and submerged depending on the location of the membrane module.

FIG. 1 illustrates a side flow type biofilm reactor, and FIG. 2 illustrates a submerged biofilm reactor.

Referring to FIGS. 1 and 2, in the sidestream biofilm reactor, the membrane module 10 is separated from the bioreactor 20, whereas in the submerged biofilm reactor, the membrane module 10 is connected to the bioreactor 20. It is provided in submerged form.

In the case of the side flow type, there is a difference in that a process of returning the sludge filtered from the membrane module 10 to the bioreactor 20 is further added, and the water purified in the bioreactor 20 is permeated through the membrane module 10. The same is true for performing the second purification.

The use of a membrane module in an activated sludge process prevents sludge and solids from passing through the membrane, allowing for complete preservation of bacterial communities, and through the second purification process, all solids can be filtered out. The advantage is that the system can be provided.

However, in the activated sludge process using such a membrane module, when the membrane-filtered suspension aqueous solution in which colloid particles and the like are dispersed is generated by a cake layer or a gel layer in which solute is accumulated on the membrane surface. Membrane fouling will reduce the overall efficiency of the process, so the membrane should be cleaned or replaced periodically.

Sludge deposits on the surface of the film, the formation of cake layers or gel layers, and the growth of microorganisms are pointed to cause fouling. In addition, particulate fouling may occur due to colloids and filtered solids, and mixed oxygen suspended solids (MLSS) dissolved oxygen and sludge retention time are also factors that affect particulate membrane fouling. Can work.

Due to the variety of factors that cause membrane fouling, the prediction of the fouling process requires a complicated calculation process, and thus there is no universal model generally accepted for the fouling mechanism.

The present invention devised to solve the above problems, the present invention provides a linear model that predicts the intermembrane pressure difference of the biofilm reaction tank, and can predict the cleaning time of the membrane module through the prediction of the intermembrane pressure difference, The purpose is to improve the accuracy of the model.

In addition, as a method for presenting a prediction model, there is another purpose of using the statistical least square method and the cyclic least square method, and examining the effectiveness of applying the RMSE to the prediction model.

The first invention of the present application, the membrane cleaning timing prediction method of the biofilm reaction tank for solving the above object, in the membrane cleaning timing prediction method of the biofilm reaction tank having an exponential relationship between the intermembrane pressure difference-time, A step 1 of deriving a linear relational expression by performing linearization of an estimated index of a time curve; Measuring the biofilm reactor operation time and the pressure difference between the two membranes; Deriving a parameter of the linear relation by applying a least square method to the measured time and the interlayer pressure difference; And four steps of deriving the membrane cleaning timing by substituting the intermembrane pressure difference when the membrane cleaning is required in the linear relational expression.

At this time, the first invention of the present application, in the method for predicting the membrane cleaning time of the biofilm reactor having an exponential relationship between the inter-membrane pressure difference-time, deriving a linear relationship by performing linearization of the estimated index of the inter-membrane pressure difference-time curve Step 1 to do; A second step of measuring a biofilm reactor operation time and an intermembrane pressure difference at a predetermined time period to derive a measurement value; Deriving a parameter of the linear relation whenever the measured value is derived by applying a cyclic least square method to the measured time and the interlayer pressure difference; And a four step of predicting the timing of membrane cleaning by applying a statistical quality control technique to the derived value of the parameter.

The second invention of the present application, the membrane cleaning timing prediction system of the biofilm reaction tank, the data measuring unit for measuring the inter-membrane pressure difference data according to the operating time from the biofilm reaction tank; A parameter calculator for calculating a parameter of a linear equation between the intermembrane pressure difference and operation time by applying a least square method to the data measured by the data measurer; And a membrane cleaning timing determination unit for predicting the membrane cleaning timing by substituting the reference value of the pre-membrane pressure difference set in the linear equation.

At this time, the second invention of the present application, the data measuring unit for measuring the inter-membrane pressure difference data according to the operating time from the biofilm reaction tank in a predetermined time period; A parameter calculator which applies a linear least square method to data measured by the data measurer and calculates a parameter of a linear equation between the intermembrane pressure difference and operation time; And a membrane cleaning timing determiner for predicting the membrane cleaning timing by applying a quality control technique to the distribution of the parameter.

According to the present invention having the configuration as described above, it is possible to present a predictive model having a normal distribution, there is an advantage that can be predicted between the membrane pressure difference between the biofilm reactor and the correct cleaning time of the membrane module.

1 is a side flow biofilm reactor according to the prior art,
2 is a biofilm reaction tank of a submerged type according to the prior art,
3a and 3b is an exemplary view showing the distribution of κ and b at daily intervals by applying a cyclic least square method to the biofilm reactor process,
4A and 4B are partial enlarged views of the time point at which the distribution of κ and b is deviated from the center line to ± 1σ, and
5 is a block diagram of a membrane cleaning timing prediction system of a biofilm reactor according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms and words used in the present specification and claims should not be construed as limited to ordinary or dictionary terms, and the inventor should appropriately interpret the concepts of the terms appropriately It should be interpreted in accordance with the meaning and concept consistent with the technical idea of the present invention based on the principle that it can be defined. Therefore, the embodiments described in this specification and the configurations shown in the drawings are merely the most preferred embodiments of the present invention and do not represent all the technical ideas of the present invention. Therefore, It is to be understood that equivalents and modifications are possible.

Intermediate pilot scale If the permeate flux is constant in the biofilm reactor process, the Trans Membrane Pressure (TMP) curve shows a typical exponential characteristic that increases steeply at the end of the filtration process between the first one and two months. As such, the sharp rise in TMP at the end of the filtration process may include a heterogeneous fouling area reduction model, a heterogeneous fouling pore reduction model, a critical suction pressure model, a percolation theory, a heterogeneous fiber bundle model, and the like. This can be explained by.

In the heterogeneous fouling area reduction model, which is one of them, the transmembrane pressure (TMP) according to the progress of the MBR processor may be expressed by Equation 1 below.

Where TMP ₀ is the initial intermembrane pressure difference when filtration time is zero before fouling starts, t means filtration time, and the unit is days. In addition, κ _t is a time-based fouling coefficient, determined by the characteristics of the MBR system, in units of l / day.

In the above equation, if the maximum allowable value of the inter-membrane pressure difference to be subjected to chemical cleaning for the membrane can be determined and the time-based characteristic value κ _t of the MBR system can be known, it is possible to calculate the time t to clean the membrane. .

In general, since the intermembrane pressure difference generated by the fouling phenomenon is 40 to 50 kPa, the maximum allowable value of the intermembrane pressure difference is preferably within 40 to 50 kPa.

If Equation 1 is applied to an MBR system of an intermediate experiment scale operating in a constant flux mode, first, the cumulative capacity passing through the filter may be expressed as Equation 2 below.

Where J is the permeate flux through the filter, A is the surface area of the membrane and t is the filtration time.

Using this, if the TMP function shown in Equation 1 is changed to a function for the cumulative capacity passing through the filter instead of the filtering time, it can be expressed as Equation 3 below.

Where κ _V represents the volume-based irreversible fouling coefficient and the unit is l / m ³ . As in Equation 1, by specifying the maximum intermembrane pressure difference required for the membrane cleaning, and given the fouling coefficient κ _V according to the system characteristics, it is possible to calculate the membrane cleaning timing by calculating the cumulative capacity V through the filter do.

In Equation 3, if it can be assumed that κ _V and the concentration of the water-soluble substance of the activated sludge solution have a proportional relationship, Equation 3 may be modified as in Equation 4 below.

Where L is the cumulative amount of organic contaminants on the membrane surface in kg, κ _L is the fouling coefficient based on organic contaminants, and the unit is l / kg. In addition, C means the density | concentration of the organic substance in the water-soluble substance of a turbid liquid.

As can be seen in Equations 1, 3 and 4, when modeling the TMP in the MBR process, TMP has the form of an exponential function rather than a linear function for t, V and L, TMP and t, V Since the distribution of and L shows a statistically nonnormal distribution, there is a problem that it is difficult to accurately predict the membrane washing interval.

In general, nonnormal distributions are used to express changes in time or space of related parameters applied in applied science and engineering. The nonnormal distribution has the characteristic that the probability density of the data has a peak at the midpoint and heavy tails when compared with the normal distribution of the same variance. The modeling system having such a nonlinear distribution has a disadvantage in that its performance is significantly lower than that of a normal distribution in performance. Therefore, for accurate prediction, it is necessary to convert the measured variables into a normal distribution or a linear model.

As can be seen in Equations 1, 3, and 4, the intermembrane pressure difference (TMP) is a lognormal distribution, which is a kind of non-Gaussian distribution having an exponential relationship. Therefore, it is difficult to predict the exact intermembrane pressure difference or to predict the time of proper washing of the membrane with such a model. To solve this problem, it is necessary to change the intermembrane pressure difference to a linearized variable or perform normalization.

In order to perform linear linearization of the estimated index of the inter-membrane pressure difference (TMP) of Equations 1, 3, and 4, natural logarithmic transformation may be attempted. In order to build a model for accurate prediction of TMP and membrane washing time, natural logarithms are taken on both sides of Equations 1, 3 and 4.

That is, taking the natural logarithm to both sides of the equations (1), (3) and (4), the exponential fouling model can be converted into a linear linear model.

For example, if the natural log is taken on both sides of Equation 1, it may be expressed as a linear function as shown in Equation 5 below.

That is, given the parameters lnTMP ₀ and κ _t , it is possible to estimate the membrane cleaning timing t by setting the limit value of TMP.

That is, if the natural log is taken on both sides of the equations (1), (3) and (4), it can be converted into a linear linear model such as the following equation (6).

That is, the modeling function of the MBR processor may be represented by a linear function having a linearity as shown in Equation 6 by taking a logarithmic function. Where y represents lnTMP, κ represents the relationship of κ _t , κ _L , κ _V , t represents the number of days of operation, and b represents the lnTMP ₀ value.

In this case, the least-square method may be used to obtain values of the parameters κ and b. The least-squares method is a statistical method for predicting the values of parameters κ and b using measured TMP values and data of days of operation t.

By predicting the values of κ and b through the least-squares method, the limiting TMP of fouling can be determined, and thus the number of days t required to clean the membrane can be estimated.

Root Mean Square Error (RMSE) can be used to validate the k and b values derived by the least-squares method. The RMSE is for measuring the difference between the value predicted by the exponential relationship and the actual measured value.

RMSE can be defined as Equation 8 below.

는 모델에 의해 예측된 TMP 값을 나타내며, n은 샘플의 수를 나타내는 것이다.Where Y _i is the actual measured TMP value,

Denotes the TMP value predicted by the model, and n denotes the number of samples.

Using the least-squares method described above, one relationship can be derived that is applicable to the biofilm reactor process. However, when the least-squares method is used, one relational expression is used throughout the entire biofilm reactor, so if the inter-membrane pressure difference is affected by an error of another measurement value, is it normal or abnormal? There is a problem that can not be determined.

In order to solve this problem, by applying the cyclic least squares method, it is possible to derive a relation for the appropriate inter-membrane pressure difference for each experimental date.

In order to apply the cyclic least-squares method, in the biofilm reactor process, the intermembrane pressure difference data should be updated periodically, and preferably updated daily. The cyclic least-squares method accumulates data updated daily and calculates κ and b values by day. Therefore, the prediction model of the intermembrane pressure difference is different at daily intervals, so that a more accurate timing of membrane cleaning can be determined.

In other words, the least-squares method uses the same values of κ and b throughout the period, but the cyclic least-squares method has the advantage of building a more accurate prediction model by updating the values of κ and b that are suitable for the cumulative data every day. have.

When the cyclic least square method is applied, statistical quality control (SQC) can be applied as a method for determining the timing of membrane cleaning. Statistical quality control techniques divide data fluctuations into meaningful and meaningless fluctuations, and find constant regularity in fluctuating data.

Figures 3a and 3b is an illustration showing the distribution of κ and b at daily intervals by applying the cyclic least square method to the biofilm reactor process, Figures 4a and 4b is the distribution of κ and b out of the center line toward ± 1σ It is a partial enlarged view at the time of becoming.

As can be seen from the points aa and bb of FIGS. 3a and 3b, 4a and 4b, as the process proceeds, κ and b approaches the centerline and moves to ± 1σ. As a result, it can be seen that b tends to increase as the process proceeds, and κ tends to decrease as the process proceeds.

In this distribution table, it can be determined that membrane cleaning is necessary at the time point at which b and κ are out of the center line and tend to shift to ± 1σ.

That is, when κ and b values updated daily by the cyclic least square method are distributed to ± 3σ based on the center line using statistical quality control technique, κ and b values converge to the center line after a certain time. After that, it goes out of the center line and goes to ± 1σ. In this case, when the κ and b values tend to deviate from the center line, it is determined that the deviation increases according to fouling, and thus it may be determined that the membrane cleaning timing is necessary.

5 is a block diagram of a membrane cleaning timing prediction system of a biofilm reactor according to an embodiment of the present invention. Referring to FIG. 5, the membrane cleaning timing prediction system of the biofilm reactor includes a data measuring unit 110, a parameter calculator 120, a membrane cleaning timing determining unit 130, and a data storage unit 140. The membrane cleaning timing prediction system of the biofilm reactor may further include a reference setting unit 150 that may set a limit inter-membrane pressure difference value when fouling occurs.

First, the data measuring unit 110 measures the TMP data according to the number of operating days from the biofilm reactor and stores it in the data storage unit 140. The data measuring unit 110 may obtain TMP data from a pressure sensor included in a biofilm reactor, or may obtain TMP data based on an external input of a user.

The parameter calculator 120 reads the measurement data from the data measuring unit 110 to derive the parameters of the TMP prediction linear model of the biofilm reactor. For example, κ and b can be derived by applying the least square method or the cyclic least square method to the measured data in the linear TMP prediction model as shown in Equation 6 above.

In this case, when the parameter is derived by applying the cyclic least square method, the data measuring unit 110 preferably updates the measurement data in a predetermined cycle, preferably, on a daily basis, and the parameter calculating unit 120 is also daily. Deriving κ and b in units and stores them in the data storage unit 140.

The membrane cleaning timing determiner 130 determines the membrane cleaning timing of the biofilm reactor using the parameter calculation result. When the parameter calculator 120 derives a parameter using the least squares method, since the TMP prediction linear model using the derived parameter is a model applied to the whole days of the process operation, fouling from the reference setter 150 is performed. This predicted marginal TMP value predicts the timing of membrane cleaning.

Alternatively, when the parameter calculator 120 derives a parameter using the cyclic least squares method, since the TMP predictive linear model using the derived parameter is a model that can be applied only to the corresponding date, the reference setter 150 may determine the parameter. By applying quality control techniques to the parameters, it is possible to predict when to clean the membrane.

As described above, although the present invention has been described by way of limited embodiments and drawings, the present invention is not limited thereto and is intended by those skilled in the art to which the present invention pertains. Various modifications and variations are possible within the scope of equivalents of the claims to be described.

10: membrane module
20: Bioreactor
110: data measuring unit
120: parameter calculation unit
130: membrane cleaning timing determination unit
140: data storage unit
150: reference setting unit

Claims (18)

In the method for predicting the membrane cleaning time of a biofilm reactor having an exponential relationship between the intermembrane pressure difference-time,
A step 1 of deriving a linear relation by performing linearization of the estimated indicators of the intermembrane pressure difference-time curve;
Deriving a measurement value by measuring a biofilm reactor operation time and an intermembrane pressure difference at a predetermined time period;
Deriving a parameter of the linear relation whenever the measured value is derived by applying a cyclic least square method to the measured time and the interlayer pressure difference; And
Four steps of deriving the membrane cleaning timing by substituting the intermembrane pressure difference when the membrane cleaning is required in the linear relational expression
Method for predicting the membrane cleaning time of a biofilm reactor comprising a.
The method of claim 1, wherein the first step is
The exponential relationship between the intermembrane pressure difference and time,

Where TMP is the intermembrane pressure difference, TMP ₀ is the first intermembrane pressure difference before filtration begins, κ _t is the time-based fouling coefficient, and t is the elapsed time of the biofilm reactor process, determined by the characteristics of the biofilm reactor system. A method for predicting the membrane cleaning time of a biofilm reactor, wherein the linear relation is derived by taking a natural log.
The method of claim 2, wherein the third step is
The linear relation

The method according to claim 1, wherein the parameters lnTMP ₀ and κ _t are derived for each time period by applying a least-squares method to the measured values of TMP and t updated at each time period.
The method of claim 3, wherein the fourth step,
A membrane cleaning timing prediction method of a biofilm reactor, characterized in that the membrane cleaning timing is derived by substituting an interlayer pressure difference of 40 to 50 kPa into the linear equation.
The method of claim 1,
Between the three and four steps,
The method of predicting the membrane cleaning time of a biofilm reactor further comprising the step of confirming the validity of the derived parameter using a root mean square error.
The method of claim 5, wherein
The root mean square error is estimated based on the measured value and the calculated value of the pressure difference between the membranes.
In the membrane cleaning timing prediction method of a biofilm reactor having an exponential relationship between the intermembrane pressure difference and the permeation capacity,
A step 1 of deriving a linear relation by performing linearization of the estimated indicators of the intertidal pressure difference-volume curve;
Deriving a measurement value by measuring a permeation capacity and a pressure difference between the membranes passing through the membrane of the biofilm reactor at predetermined time periods;
Deriving a parameter of the linear relation whenever the measured value is derived by applying the cyclic least square method to the measured permeation capacity and the intermembrane pressure difference; And
Four steps of deriving the membrane cleaning timing by substituting the intermembrane pressure difference when the membrane cleaning is required in the linear relational expression
Method for predicting the membrane cleaning time of a biofilm reactor comprising a.
The method of claim 7, wherein the first step,
The exponential relationship between intermembrane pressure difference and permeation capacity

(Wherein TMP is the intermembrane pressure difference, TMP ₀ is the first intermembrane pressure difference before filtration starts, κ _V is the volume-based fouling coefficient determined by the characteristics of the biofilm reactor system, and V is the permeate capacity through the membrane). A method for predicting the membrane cleaning time of a biofilm reactor, wherein the linear relation is derived by taking a natural log.
The method of claim 7, wherein the three steps,
The linear relation

The method according to claim 1, wherein the parameters lnTMP ₀ and κ _V are derived on the basis of the time period by applying the least-square method to the measured values of TMP and v updated at each time period.
The method of claim 9, wherein the fourth step,

A method for predicting the membrane cleaning time of a biofilm reactor, wherein the membrane cleaning timing is derived using (J is the permeate flux passing through the membrane, A is the surface area of the membrane, and t is the biofilm reactor operation time).
In the method for predicting the membrane cleaning time of a biofilm reactor having an exponential relationship between the intermembrane pressure difference-time,
A step 1 of deriving a linear relation by performing linearization of the estimated indicators of the intermembrane pressure difference-time curve;
Deriving a measurement value by measuring a biofilm reactor operation time and an intermembrane pressure difference at a predetermined time period;
Deriving a parameter of the linear relation whenever the measured value is derived by applying a cyclic least square method to the measured time and the interlayer pressure difference; And
Four steps to predict the timing of membrane cleaning by applying statistical quality control techniques to the derived values of the parameters
Method for predicting membrane cleaning time of a biofilm reactor comprising.
The method of claim 11, wherein the first step,
The exponential relationship between the intermembrane pressure difference and time,

Where TMP is the intermembrane pressure difference, TMP ₀ is the first intermembrane pressure difference before filtration begins, κ _t is the time-based fouling coefficient, and t is the elapsed time of the biofilm reactor process, determined by the characteristics of the biofilm reactor system. A method for predicting the membrane cleaning time of a biofilm reactor, wherein the linear relation is derived by taking a natural log.
The method of claim 12, wherein the third step,
The linear relation

The method for predicting membrane cleaning timing of a biofilm reactor according to claim 1, wherein the parameters lnTMP ₀ and κ _t are derived on a daily basis by applying a least square method to the measured values of TMP and t updated daily.
The method of claim 13, wherein the four steps,
And a time when the distribution of the parameter converges to a center line and then moves out of the center line toward ± 1 s as a membrane cleaning timing.
A data measuring unit configured to measure intermembrane pressure difference data according to an operation time from a biofilm reaction tank at predetermined time periods;
A parameter calculator which applies a cyclic least square method to the data measured by the data measurer and calculates a parameter of a linear equation between the interlude pressure difference and operation time each time the data is measured; And
Membrane cleaning timing determining unit for predicting the membrane cleaning timing by substituting the reference value of the pre-membrane pressure difference set in the linear equation
Membrane cleaning timing prediction system of a biofilm reactor comprising a.
A data measuring unit configured to measure intermembrane pressure difference data according to an operation time from a biofilm reaction tank at predetermined time periods;
A parameter calculator which applies a cyclic least square method to the data measured by the data measurer and calculates a parameter of a linear equation between the interlude pressure difference and operation time each time the data is measured; And
Membrane cleaning time determination unit for predicting the membrane cleaning time by applying the quality control technique to the distribution of the parameters
Membrane cleaning timing prediction system of a biofilm reactor comprising a.
17. The method according to claim 15 or 16,
The linear equation is the membrane cleaning timing prediction system of the biofilm reactor, characterized in that the following equation.

[Mathematical Expression]

(Where TMP is the intermembrane pressure difference, κ _t is the time-based fouling coefficient determined by the characteristics of the biofilm reactor system, and t is the number of days of operation).
17. The method of claim 16,
The membrane cleaning timing determining unit predicts a time when the distribution of the parameter converges to a center line and then moves out of the center line toward ± 1σ as the membrane cleaning timing. Timing prediction system.

Images