EP2644772A1 - A method for controlling a finishing device of a fiber web - Google Patents

Abstract

Method for controlling a finishing device of a fiber web at a transition stage of the process for the break recovery a break recovery model is created based on measurement results by creating a repair profile and/or optimizing the measuring by several measurement results.

Description

• The invention relates generally to a method for controlling a finishing device of a fiber web at a transition stage of the process, in particular for controlling a calender at splicing stage. More especially the invention relates to a method according to the features of the preamble part of claim 1.
• In connection of finishing devices of a fiber web, for example in connection of calender in calendering a paper or broad web, the properties of fiber web are constantly monitored by means of on-line measurements. The measurements are conducted in the cross-direction (CD) of fiber web in order to produce the profile of the measured property in the cross direction of the fiber web. Typically the measurements are performed by means of measuring apparatuses, in which a moving fiber web is measured by means of measuring sensors moving back and forth in its cross direction. The properties to be measured may include for example moisture, caliper, basis weight, ash content, color, opacity, brightness, gloss, or smoothness of the web. The results obtained from the measuring sensors are used not only for monitoring the properties of the fiber web, but also for controlling the finishing devices of the fiber web. The measurement results are transmitted to a control unit, in which they are utilized to determine control signals for profiling apparatuses belonging to the finishing process of fiber web and affecting said property in the cross direction of the fiber web. Each of these profiling apparatuses contains one or several actuators affecting a point corresponding to their location in the cross direction of the web. The control profile of the profiling device typically comprises the control signals of the actuators relating thereto. When controlling cross direction profiles, the processing of signals is typically performed by processing information in profile form. For each variable to be measured an error profile is determined, the error profile being the deviation between the profile formed on the basis of the measurement results and the target profile set for the variable, said error profile describing the error in the adjustment. The purpose of the control is to keep the process as accurately as possible in a state complying with the targets determined for the process. By means of the error profile the control unit forms control commands for one or several profiling devices or actuators that affect the process and bring about a change therein complying with the control commands.
• One problem in the finishing devices of a fiber web is regularly occurring disturbances in the operating stages i.e. transition stages deviating from the normal run. The disturbances are typically similar in similar situations and they produce defects in the web that is being manufactured. As a result of the defects the target quality of the web is not reached and the product produced in the process cannot be delivered to a client, but it is treated as broke. This is not cost-effective. The transition stages in which the above-mentioned recurrent errors occur include for example a disturbance in the process, a change in a set value relating to the process, starting up of the process or its parts or deceleration before stopping the process. For example after a break or in connection with splicing, when the process has been started again, the quality of the product does not typically correspond to the target values set for the product, but the target values are reached only after a while from starting the production. The control unit of the process, the automation systems and the actuators control the process during the entire transition stage, but it takes time to reach acceptable product quality. There have been attempts to shorten the time passed for reaching the target quality in various ways, for example by running the process manually. In the manual run the operator can correct the quality of the product by changing the position of the actuators in a way that deviates from the function of the automatic control.
• In prior art publication WO 2008/065252 a method and a system for controlling the manufacturing or finishing process of a fiber web is disclosed which method and system are is based on the idea that empirical information, such as correction profiles are utilized for forming control signals to be transmitted to actuators at a transition stage of a manufacturing or finishing process of a fiber web, which correction profiles can be updated by means of an error profile of a web property formed in a control unit. The correction profiles are determined in the control unit that calculates new control signals for the actuators. The determined correction pro files are stored in the memory means of the control unit so that they form correction profile series. One correction profile series comprises successive correction profiles determined during one transition stage of the process. Each one of the correction profiles in the series is connected to the progress of the transition stage, i.e. to one or several calculations of the control signal. Control signals and correction profiles series determined in earlier corresponding transition stages and stored in memory means are utilized for producing new control signals. Error profiles determined by means of measurements attained from the process and target values of the process are also utilized. When the transition stage begins, one of the correction profile series determined in a corresponding earlier transition stage is selected and used in the calculation. Individual correction profiles of the selected correction profile series are used for forming the control signals so that on the basis of an individual old correction profile selected from the series and the determined error profile a corrected error profile is formed, which is used for forming a control signal. The determined error profiles are also utilized for updating a correction profile used in the previous calculation and stored in the memory means. Thus, each correction profile contains empirical information for the next calculation cycle of the control signal, by means of which the control unit is capable of performing the necessary correction for compensating the effect of the disturbance beforehand in the calculation of the control signals.
• Calendering is generally carried out in order to improve the properties, like smoothness and gloss, of a web-like material such as a paper or board web. In calendering the web is passed into a nip, i.e. calendering nip, formed between rolls that are pressed against each other or a treatment zone between a roll and a belt or between two belts, in which nip the web becomes deformed as by the action of temperature, moisture and nip pressure. In the calender the nips are formed between a smooth-surfaced press roll such as metal roll and a roll coated with resilient material such as a polymer roll. The resilient-surfaced roll adjusts itself to the forms of the web surface and presses the opposite side of the web evenly against the smooth-surfaced press roll. Multiroll calenders have usually a roll whose profile can be adjusted , for example a variable-crown calender roll containing inside itself one or several profiling members, such as loading elements affecting the shell of the roll radially in the direction of the axis of the roll. The loading elements are typically hydraulic pistons which are pressed against the shell of the roll to form the desired profile for the load, i.e. nip load transmitted via the roll to the nip and further to the paper web to be calendered. Thus, it is at the same time possible to compensate the change in the profile caused by the deflection of the roll. The number of loading elements depends on the width of the roll, and they are typically positioned at intervals of 10 to 20 cm in the direction of the axis of the roll. The loading elements can be controlled separately. The controlling takes place by controlling the oil pressure of the loading elements by means of the control system. By controlling the profiling actuator manually at the transition stage, it is possible to reduce the effect of the disturbance to a certain extent. However, the manual method is always very dependent on the skills and experience of the operator. In practice, it has been discovered that by keeping the automatic control switched off during such a production stage until the direction of the error development changes, the time passed after the operating stage for the recovery of the production is shortened approximately 30%. By predicting the error after or before the operating stage it is possible to attain even better results. However, this result is completely insufficient economically.
• The purpose of the present invention is to introduce such a method for controlling a finishing device of a fiber web at a transition stage to avoid the aforementioned problems and enable the control of the process in such a manner that it is possible to minimize the amount of product to be treated as broke or reject.
• To attain the purposes mentioned above and those that will be presented later, the method according to the invention is primarily characterized in what will be presented in the characterizing part of the independent claim 1.
• The other, dependent claims will present some preferred embodiments of the invention.
• In the prior art methods for controlling a finishing device of a fiber web at a transition stages in many cases first measurement has to be rejected automatically even though the measurement might be valid or a failure is not shown until the second measurement in situations relating to load mode changes/change for example in the calender in changes nip closed - nip with relief - nip under load. In calenders the first measurement is usually obtained from the measuring frame 20 s - 4 min after the calender is loaded thus forming a break recovery model and performing well-timed repairs are difficult and furthermore, the measuring frame or measuring device must be stopped at intervals during the break recovery for cleaning or for service, whereby break recovery is continued often for several minutes incorrectly. The break recovery model is updated based on one valid measurement or the model grows too big in failure situations and thus restoring the model operational requires several break recoveries. Break recovery models become spiky without filtering and spiky repair profile is filtered nonexistent in the optimisation without effect.
• According to one advantageous embodiment of the invention in the method for controlling a finishing device of a fiber web at a transition stage for the break recovery sequence an automatic rejection of measurement result is used at the beginning such that in load change situations, for example in a calender after the nip has been loaded, measurement profiles are rejected in less than 60 seconds after loading. After this the next measurement is waited and if the time from the previous measurement (valid or invalid) is more than set time interval, for example 60 seconds, the step of the break recovery sequence is released to be time-based. In the time-based sequence, the sequence is stepped according to a set measurement update interval, for example 45-second measurement update interval. Stepping is continued time-based forwards until new measurements are obtained. When a new measurement is obtained, the stepping of the break recovery sequence is continued according to the measurement update interval. If an invalid measurement is obtained during the break recovery sequence, the break recovery sequence is stepped normally forwards. If the time for automatic rejection, for example 60 seconds, has passed from the measurement previous to the measurement, the sequence starts to be stepped time-basely. In creating the break recovery model the repair profile is only updated if during a step, the measurements are valid (measurements during the time-based stepping are not considered valid). A maximum profiling window is set for the repair profile and a new repair profile is validated according to the maximum profiling. If the new repair profile is not valid, the repair profile is not updated. Also the repair profile learning is filtered and difference data from the controller is filtered such that the shape of the roll is visible. Otherwise, possible poor profile points characteristic of each paper machine will build up.
• Another advantageous embodiment of the invention in the method for controlling a finishing device of a fiber web at a transition stage relates especially to optimizing the measuring by using measurement technology that is based on measurements on the whole width of the web (parent roll) during each revolution of the roll using for example the intelligent measurement roll by the applicant that is based on measuring with pressure sensitive film that is called by trade name iRoll. The measuring roll receives a measurement result every roll revolution. By this results of several measurements during deceleration and acceleration is received and thus reliable information of quality is received and broke amount can be minimized.
• One advantageous feature of the invention in the method for controlling a finishing device of a fiber web at a transition stage relates especially to minimizing amount of broke web during splicing of the web, in particularly in a calender. The web produced during the time the running speed is outside the quality based limits while decelerating the running speed of a finishing device to the splicing speed and accelerating the running speed back to the production speed is typically treated as broke thus the time used for the decelerating stage and the accelerating stage outside the quality based speed limits should be as short as possible. According to the invention this is achieved by optimizing the deceleration and acceleration such that the deceleration and acceleration is faster when outside the quality based speed limits. By this the amount of broke web is smaller and also the capacity is increased.
• In this description and in the claims the concept of a transition stage of a process refers to a recognizable operating stage deviating from the normal run of the process. At this stage the process is running and during the process regularly occurring errors are detected in the CD profile of a certain property of the web. Such transition stages include for example starting up of the process or its parts or deceleration of the functions of the process before stopping the process. Errors can be caused for example by the structural properties of the actuators or parts of the process, for example felts or wires used in the manufacturing or finishing line of paper or paperboard. The concept of a regularly occurring error refers to an error occurring either during the entire operating stage or at regular intervals, which error can be seen in the CD profile of a web property measured from the web. In this connecting the calender roll refers to rolls that form the calendering nip.
• In the following, the invention will be described in more detail with reference to the appended drawings, in which
• figure 1 shows schematically one example of an advantageous feature of the invention and
• figure 2 shows schematically another example of an advantageous feature of the invention.
• An example of a quality control based model is explained in the following which is a static model example for a calender with the control variables of (m (i,t)) nip load, roll surface temperature, dwell time, grade, moisturising, drying, steaming and their correlation to output variables i.e. quality variables (y(j,t)): gloss, smoothness, calibre, bulk, end-moisture, blackening, opacity and porosity. In this model measurements y(j)_lab most important for production are performed in a laboratory and only part of measurements can be measured in real-time directly from the process. The model uses laboratory samples and limited on-line measurements based on the following mathematical calculations: $$\mathrm{y}\left(\mathrm{j}\mathrm{t}\right)\mathrm{enn}\mathrm{2}\mathrm{=}\mathrm{f}\left(\mathrm{m}\left(\mathrm{i}\mathrm{t}\right)\mathrm{,}\mathrm{j}\right)\mathrm{=}\mathrm{y}\left(\mathrm{j}\right)\mathrm{lab}\mathrm{1}\mathrm{+}\mathrm{b}\left(\mathrm{j}\mathrm{i}\right)\mathrm{*}\mathrm{m}\left(\mathrm{i}\mathrm{t}\right)\mathrm{-}\mathrm{e}\left(\mathrm{j}\right)\mathrm{1}\mathrm{,}$$

  

  
  in which $$\mathrm{e}\left(\mathrm{j}\right)\mathrm{1}\mathrm{=}\mathrm{y}\left(\mathrm{j}\right)\mathrm{enn}\mathrm{1}\mathrm{-}\mathrm{y}\left(\mathrm{j}\right)\mathrm{lab}\mathrm{1}$$

  

  
  and $$\mathrm{y}\left(\mathrm{j}\mathrm{t}\right)\mathrm{enn}\mathrm{3}\mathrm{=}\mathrm{f}\left(\mathrm{m}\left(\mathrm{i}\mathrm{t}\right)\mathrm{,}\mathrm{j}\right)\mathrm{=}\mathrm{y}\left(\mathrm{j}\right)\mathrm{lab}\mathrm{2}\mathrm{+}\mathrm{b}\left(\mathrm{j}\mathrm{i}\right)\mathrm{*}\mathrm{mit}\left(\mathrm{i}\mathrm{t}\right)\mathrm{-}\mathrm{e}\left(\mathrm{j}\right)\mathrm{2}\mathrm{,}$$

  

  
  in which $$\mathrm{e}\left(\mathrm{j}\right)2\mathrm{=}\mathrm{y}\left(\mathrm{j}\right)\mathrm{enn}2\mathrm{-}\mathrm{y}\left(\mathrm{j}\right)\mathrm{lab}2.$$

  
• The mathematical model is y(j,t)_enn = f(m(i,t),j) , where j is desired quality variable, i is specific control variable and f (m(i,t),j) is function or model which predicts the effect of used control variables to the desired quality variable. When running the process, it is possible to calculate a prediction for desired quality variables, when changes are made to the control variables. Part of quality variables can also be measured on-line and they can be adjusted as direct feedback with selected control variables, but as the control variables have cross-effects to other quality variables, the optimal adjustment requires considering the cross-effects. When the laboratory measurement y(j)_lab is obtained, it is compared to a prediction made by means of control variables before the laboratory measurements y(j)_lab - y(j,t_lab)_enn = e(j) and, by means of the difference, the reliability of the next prediction is corrected. The updating moment of prediction parameters is (tlab). In this example the prediction can be improved by adding run-time failure effects to it, if their effect can be predicted on the quality variable in question and m(i,t) is change at time t compared to the value at the time of laboratory measurement sample m(i, tlab).
• When limited on-line measurements are used is good to have at least one MD- and CD-directional measurement in real-time. The least costly MD measurement is web moisture and temperature and in CD direction iRoll and web temperature. Calculated real-time measuring data is obtained from MD calibre and basis weight on the winder when MD moisture is known. Real-time measurements give a possibility to evaluate the effect of process failures between laboratory measurements on the quality variables being predicted. By means of iRoll, CD temperature and MD moisture measurements, it is also possible to make a real-time CD basis weight, CD gloss, CD moisture estimate.
• Optimisation is choosing the best alternative with desired valuations and given boundary conditions/limitations. The desired valuations are expressed with an object function which is generally a so-called cost function: $$\mathrm{Min}\left(\mathrm{a}\mathrm{\times }\mathrm{quality\; deviations}\mathrm{+}\mathrm{b}\mathrm{\times }\mathrm{energy}\mathrm{+}\mathrm{c}\mathrm{\times }\mathrm{shutdowns}\right)\mathrm{.}$$

  
• The boundary conditions are limitations dependent on the device, the greatest quality deviations allowed etc. when the process is controlled manually, the operator optimises the process heuristically based on available data and experience. If the process is controlled optimally by means of automation, the above-described object and limitation functions have to be formed and already in multi-variable controls, the control choices have to be evaluated optimisation-wise. By estimating in real time measurement and based on measurement data of laboratory measurements can be combined in the models and needed corrections in control can be made in direct controlling of the finishing device.
• As shown in figure 1 the web produced during the time, X-axis, the running speed, in prior art as indicated by line D/A13 and in accordance with the advantageous feature of the invention as indicated by line D/A12, is outside the quality based limits, below line Z , while decelerating the running speed of a finishing device to the splicing speed, Y-axis, and accelerating the running speed back to the production speed is typically treated as broke thus the time used for the decelerating stage and the accelerating stage outside the quality based speed limits should be as short as possible. According to the invention this is achieved by optimizing the deceleration and acceleration such that the decoration and acceleration is faster when outside the quality based speed limits. By this the amount of broke web is smaller, compare arrows B12, indicates arrangement according to an advantageous feature of the invention, and B13 indicates arrangement according to prior art, and also the capacity is increased, indicted by arrow C.
• As shown in figure 2 according to prior art while decelerating and accelerating the measurement data is updated in long intervals, measurement points M13, and especially during the critical stages of changing the acceleration rate and control the measurement data is updated only for example in 30 to 40 seconds intervals. This leads to situation in which deceleration is run without real time measurement data of profile. According to an advantageous feature of the invention by using measurement technology that is based on measurements on at least 60% of the whole width of the web (parent roll) during each revolution of the roll using for example the intelligent measurement roll by the applicant that is based on measuring with pressure sensitive film that is called by trade name iRoll. The measuring roll receives a measurement result, measurement point M12, every roll revolution. By this results of several measurements during deceleration and acceleration is received and thus reliable information of quality is received and broke amount can be minimized. In the figure in Y-axis is splicing speed and in X-axis time and further arrows T indicated top quality web and arrow R broke.
• One transition stage of a finishing device of a fiber web machine is the break recovery stage of a finishing device and for example in connection with a calender break recovery disturbances due to calender change from loading to relief occur and thus measurement during that stage can be valid or load change may result a measurement difference. In cases first measurement after break may not be valid due to load change, due to wrong stage in synchronizing or due to need of cleaning the measurement frame. Measurement profile may also be spiky due to the fiber web machine and spikes may be filtered in optimization. According to advantageous features of the invention measurement result obtained too quickly after break are rejected and only after a determined time interval obtained measurement results are considered as valid and stepwise approach is taken when new measurement results are obtained such that the correction profile is focused to right phase. Measuring is continued by determined steps if no valid results are obtained and profile update is done only if two valid measurement results are received. In cases spiky measurement profile is received in CD-direction of the fiber web the results are mapped to correct the web at right location be checking the measurement frame location in relation to the device for example checking the measurement results of each end in CD-direction. In this example the form of the roll in CD-direction is obtained and spikiness can be removed by estimating with self-learning programming predictive method and filtering the data such that irrelevant data is omitted. Learning time is also dependent on length of the break due to cooling of the rolls.

Claims (14)

1. Method for controlling a finishing device of a fiber web at a transition stage of the process, characterized in, that for controlling the finishing device of a fiber web at the transition stage for the break recovery a break recovery model is created based on measurement results by creating a repair profile and/or optimizing the measuring by several measurement results.
2. Method according to claim 1, characterized in, that for controlling a finishing device of a fiber web at a transition stage for the break recovery model and the repair profile an automatic rejection of measurement result is used at beginning and next measurement is waited until the time from the previous measurement is more than set time interval and then a step of a break recovery sequence is released to be time-based.
3. Method according to claim 2, characterized in, that in the time-based sequence, the sequence is stepped according to a set measurement update interval and stepping is continued time-based forwards until new measurements are obtained.
4. Method according to claim 3, characterized in, that stepping of the break recovery sequence is continued according to the measurement update interval after new measurement is obtained.
5. Method according to any of claims 2 - 4, characterized in, that in the method when an invalid measurement is obtained during the break recovery sequence, the break recovery sequence is stepped forwards.
6. Method according to any of claims 2 - 5, characterized in, that in the method when the time for automatic rejection has passed from the measurement previous to the measurement, the sequence starts to be stepped time-basely.
7. Method according to any of claims 1 - 6, characterized in, that in creating the break recovery model the repair profile is only updated if during a step, the measurements are valid and a maximum profiling window is set for the repair profile and a new repair profile is validated according to the maximum profiling.
8. Method according to any of claims 1 - 7, characterized in, that in the method the repair profile learning is filtered and difference data from the controller is filtered.
9. Method according to claim 1, characterized in, that for controlling a finishing device of a fiber web at a transition stage for the break recovery model for optimizing the measuring measurement is based on measurements on the whole width of the web during each revolution of the parent roll.
10. Method according to claim 1, characterized in, that measurement is based on measurements with an intelligent measurement roll advantageously is based on measuring with pressure sensitive film.
11. Method according to claim 1, characterized in, that several measurements during transition stage are received and thus reliable information of quality is received and broke amount can be minimized.
12. Method according to claim 1, characterized in, that at a transition stage relates for minimizing amount of broke web during splicing of the web deceleration and acceleration such that the deceleration and acceleration is faster when outside the quality based speed limits.
13. Method according to claim 1, characterized in, that in the method the properties to be measured comprise moisture and/or caliper and/or basis weight and/or ash content and/or color and/or opacity and/or brightness and/or gloss and/or smoothness of the web.
14. Method according to claim 1, characterized in, that in the method at least one of the measurement is a real-time measurement.

Images