[0002] The invention further relates to an apparatus for controlling the operation of stock preparation of a paper machine, the stock preparation being configured to produce machine stock to be fed to the short circulation of the paper machine either from one or more component stocks by blending them with each other, and the stock preparation including a plurality of succes- sive blending points, where the component stocks are blended with each other, a second raw material of the machine stock is added to the stock and/or the stock is diluted by blending dilution water with the stock, the apparatus be- ing configured to adjust the flow and/or consistency of one or more stocks ar- riving at a blending point and/or the concentration of the second raw material of the machine stock in the stock arriving at a blending point and/or to adjust the flow and consistency of the stock leaving a blending point and/or the con- centration of the second raw material of the machine stock in the stock leaving a blending point.

[0003] The invention also relates to a method of controlling the op- eration of stock preparation of a paper machine, the stock preparation being configured to produce machine stock to be fed to the short circulation of the paper machine either from one or more component stocks by blending them with each other, and in which stock preparation a second raw material of the machine stock is added to the stock and/or the stock is diluted by blending di- lution water with the stock, and in which method the flow and consistency of

the stock flowing forward in the dosing line of the stock preparation and/or the concentration of the second raw material of the machine stock in the stock are adjusted.

[0004] In papermaking, the measured basis weight or variables de- rived from it, such as the air-dry or completely dry basis weight, of a paper web being made are adjusted by controlling the dosing of the paper stock or paper pulp being transferred to the short circulation of the paper machine from the stock preparation of the paper machine. The stock fed into the short circulation of the paper machine is typically called machine stock. Since the quality and amount of machine stock produced cannot be made so even that the stock could be led from the manufacturing equipment directly to the paper machine, the stock preparation comprises a plurality of different storage and intermedi- ate chests. The various component stocks included in the machine stock, i. e. stocks containing different kinds of fibers, constitute the first raw material of the machine stock, i. e. the fibrous raw material of the machine stock. Different fill- ers, additives or chemicals added to the machine stock or the component stocks constitute the second raw material of the machine stock. These different fillers, additives and chemicals are used to improve the quality and printability of finished paper or the operability of the manufacturing process. Typically, the component stocks, fillers, additives and chemicals are stored in large storage chests. The composition of the machine stock conveyed to the paper machine is adjusted in a dosing system at the stock preparation, where the different stock components included in the stock are blended with each other both in a pipe leading to the blending chest and in the blending chest itself, from where the stock is conveyed to the machine chest and from there further to the short circulation of the paper machine. The consistency of the machine stock con- veyed to the short circulation is typically kept at three percent. Since the con- sistency of the component stocks stored in storage chests is usually 10 to 14%, and the consistency of repulped stock is usually about 5%, the consis- tency of the different component stocks and, if need be, that of the blended stock are diluted by addition of water, which is typically white water separated from the short circulation of the paper machine. Accordingly, the consistency of the stock to be fed into the paper machine is adjusted by changing the amount of dilution water fed into the stock, i. e. the adjustment of the consistency of the stock always relates to the addition of dilution water to the stock in a suitable ratio to the amount and consistency of the stock.

[0005] The basis weight of the paper web to be made is adjusted by changing the fiber flow conveyed to the paper machine. In practice, the basis weight is adjusted by changing the flow of machine stock. Since the basis weight adjustment is unaware of future variations in stock consistency, consis- tency variations can be eliminated for instance by including, in the machine stock flow request, an additional specification'at 3% consistency'relating to the consistency of the stock to be fed, i. e. the desired consistency of the ma- chine stock to be fed is 3%. If the measured consistency deviates from the tar- get, the flow target is amended respectively. The desired fiber flow is thus con- veyed to the paper machine. The basis weight adjustment requests for the necessary amount of fiber flow or machine stock flow from the machine chest in the stock preparation, the intent being to keep a constant amount of stock therein at all times. The change in the machine stock flow caused by a change in the basis weight, i. e. a flow disturbance, travels from the paper machine to- wards the storage towers for the component stocks, the flow disturbance being strengthened further by the action of the adjustment of the surface level in each intermediate chest in the dosing line. Because the flow disturbances are strong and rapid, the consistency adjustments are unable to keep up, causing consistency disturbances that proceed along with the stock flow towards the paper machine. Because of the large volume of the intermediate chests and the significant length of the dosing line, the process involves long delays, wherefore the stock preparation adjustment is extremely sensitive to variations both in the consistency of the component stocks and in the concentrations of fillers, additives and chemicals, which, in turn, easily lead to retention variation in the wire section of the paper machine. Retention variation in the wire section also causes changes in the ash content and basis weight of the paper web.

Accordingly, the flow disturbance caused by a change in the basis weight first proceeds as a flow disturbance from the machine chest through the dosing chests towards the stock towers and returns as a consistency disturbance through the dilution steps in the dosing line to the machine stock and further all the way to the basis weight of the paper. The basis weight is measured at the dry end of the paper machine immediately before the web is reeled into a ma- chine roll, whereby a basis weight error detected in the measurement causes a new change, i. e. a flow disturbance, by changing the machine stock flow. As a result, a state of vibration, which is difficult to manage, is created, during which paper or board having the wrong basis weight and ash content is produced.

This vibration also causes other disturbances in the operation of the process via dilution lines, for example.

[0006] The above-described dosing solution based on the blending capacity of the machine chest and the blending chest is not the only usable so- lution. Component stock dosing and blending can also be solved in other ways. For example, the machine chest and the blending chest may be two successive blending chests and the machine chest as a third chest, whereby blending is believed to be under still better control. If the consistency and other properties of the component stocks are well controlled, one chest may be suffi- cient. In new solutions, the aim is to blend the component stocks with each other in a separate blending device in the short circulation, whereby the proc- ess does not include any blending chest or machine chest.

[0007] At present, stock preparation dosing is controlled by adjust- ing the surface levels of the different stock chests and the consistencies and flow rates of the stock flows at different points of the process with unit control- lers based on feed forward coupling, examples of which are the method for regulating the surface level and the consistency in a stock chest for a compo- nent stock disclosed in US 6,210, 529, and the method for regulating the basis weight of paper or board by dosing component stocks disclosed in US 6,203, 667, both methods utilizing feed forward coupling to adjust the process.

However, the use of feed forward coupling in the adjustment is problematic, since when feed forward coupling is used, the action of adjustment changes on the process part succeeding the controller cannot be taken into account.

[0008] The object of the present invention is to provide a new type of solution for controlling the operation of stock preparation.

[0009] The method of the invention is characterized by determining the consistency of one or more stocks arriving at a blending point or the con- centration of the second raw material of the machine stock in the stock arriving at a blending point and by determining the consistency of the stock leaving a blending point or the concentration of the second raw material of the machine stock in the stock leaving a blending point, by determining the flow of one or more stocks arriving at the blending point and the flow of the stock leaving the blending point, by determining a consistency prediction for the consistency of one or more stocks arriving at the blending point or a concentration prediction of the second raw material of the machine stock in the stock arriving at a blending point, by determining a flow prediction for the flow of the stock leaving

the blending point, by determining a consistency target for the consistency of one or more stocks arriving at the blending point or a target concentration of the second raw material of the machine stock in the stock arriving at the blend- ing point and/or by determining a consistency target of the consistency of the stock leaving the blending point or the target concentration of the second raw material of the machine stock in the stock leaving the blending point, by de- termining a flow target for the flow of one or more stocks arriving at the blend- ing point and/or a flow target for the flow of the stock leaving the blending point, and by adjusting the flow and/or consistency of one or more stocks arriv- ing at the blending point and/or the concentration of the second raw material of the machine stock in the stock arriving at the blending point based on the flow prediction of the stock leaving the blending point and/or the consistency prediction of one or more stocks arriving at the blending point and/or the concentration prediction of the second raw material of the machine stock in such a manner that the flow of one or more stocks arriving at the blending point follows the determined flow target and/or the consistency follows the determined consistency target and/or the concentration of the second raw material of the machine stock in the stock follows the determined target concentration and/or by adjusting the flow and consistency of the stock leaving the blending point and/or the concentration of the second raw material of the machine stock in the stock leaving the blending point based on the flow prediction of the stock leaving the blending point and/or the consistency prediction of one or more stocks arriving at the blending point and/or the concentration prediction of the second raw material of the machine stock in such a manner that the flow of the stock leaving the blending point follows the determined flow target and the consistency follows the determined consistency target and/or the concentration of the second raw material of the machine stock in the stock follows the determined target concentration.

[0010] The apparatus of the invention is characterized in that the apparatus is configured to determine the consistency of one or more stocks ar- riving at a blending point or the concentration of the second raw material of the machine stock in the stock arriving at a blending point and to determine the consistency of the stock leaving a blending point or the concentration of the second raw material of the machine stock in the stock leaving a blending point, to determine the flow of one or more stocks arriving at the blending point and the flow of the stock leaving the blending point, to determine a consistency

prediction of the consistency of one or more stocks arriving at the blending point or a concentration prediction of the second raw material of the machine stock in the stock arriving at the blending point, to determine a flow prediction for the flow of the stock leaving the blending point, to determine a consistency target for the consistency of one or more stocks arriving at the blending point or a target concentration of the second raw material of the machine stock in the stock arriving at the blending point and/or to determine a consistency tar- get for the consistency of the stock leaving the blending point or a target con- centration for the second raw material of the machine stock in the stock leaving the blending point, to determine a flow target for the flow of one or more stocks arriving at the blending point and/or a flow target for the flow of the stock leav- ing the blending point and to adjust the flow and/or consistency of one or more stocks arriving at the blending point and/or the concentration of the second raw material of the machine stock in the stock arriving at the blending point based on the flow prediction of the stock leaving the blending point and/or the consis- tency prediction of one or more stocks arriving at the blending point and/or the concentration prediction of the second raw material of the machine stock in such a manner that the flow of one or more stocks arriving at the blending point follows the determined flow target and/or the consistency follows the de- termined consistency target and/or the concentration of the second raw mate- rial of the machine stock in the stock follows the determined target concentra- tion and/or to adjust the flow and consistency of the stock leaving the blending point and/or the concentration of the second raw material of the machine stock in the stock leaving the blending point based on the flow prediction of the stock leaving the blending point and/or the consistency prediction of one or more stocks arriving at the blending point and/or the concentration prediction of the second raw material of the machine stock in such a manner that the flow of the stock leaving the blending point follows the determined flow target and the consistency follows the determined consistency target and/or the concentration of the second raw material of the machine stock in the stock follows the deter- mined target concentration.

[0011] The method of the invention for adjusting the flow and con- sistency of the stock leaving the blending point and/or the concentration of the second raw material of the machine stock in the stock is further characterized by determining flow data and a flow prediction for the machine stock, determin- ing flow data and a flow prediction for one or more component stocks, transfer-

ring the flow data and flow prediction of a component stock along the dosing line of the stock preparation backwards in such a manner that adjustments controlling the stock flows forward along the dosing line utilize predicted future flow changes, determining consistency data and a consistency prediction of one or more component stocks and/or the concentration of the second raw material of the machine stock in the stock and a concentration prediction in the stock, and transferring the consistency data and consistency prediction of a component stock and/or the concentration of the second raw material of the machine stock in the stock and a concentration prediction in the stock forward along the dosing line in the stock preparation in such a manner that the ad- justments controlling the consistency of the stock or the concentration of the second raw material of the machine stock in the stock utilize the predicted fu- ture consistency changes or the changes in the concentration of the second raw material of the machine stock in the stock.

[0012] An essential idea of the invention is to control the operation of the stock preparation of a paper machine, the stock preparation being adapted to produce machine stock to be fed into the short circulation of the paper machine from either one component stock or several component stocks by blending them with each other and comprising a plurality of successive blending points where the component stocks are blended with each other, the second raw material of the machine stock is added to the stock and/or the stock is diluted by mixing dilution water to the stock, by adjusting the flow and/or consistency of one or more stocks arriving at a blending point and/or the concentration of the second raw material of the machine stock in the stock arriving at a blending point and/or by adjusting the flow and consistency of the stock leaving a blending point and/or the concentration of the second raw ma- terial of the machine stock in the stock leaving a blending point. The essential idea comprises determining the consistency of one or more stocks arriving at the blending point or the concentration of the second raw material of the ma- chine stock in the stock arriving at the blending point, determining the consis- tency of the stock leaving the blending point or the concentration of the second raw material of the machine stock in the stock leaving the blending point, de- termining the flow of one or more stocks arriving at the blending point and the flow of the stock leaving the blending point. The essential idea further com- prises determining a consistency prediction for the consistency of one or more stocks arriving at the blending point or a prediction for the concentration of the

second raw material of the machine stock in the stock arriving at the blending point, determining a flow prediction for the flow of the stock leaving the blend- ing point, determining a consistency target for the consistency of one or more stocks arriving at the blending point or a target concentration of the second raw material of the machine stock in the stock arriving at the blending point and/or determining a consistency target for the consistency of the stock leav- ing the blending point or the target concentration of the second raw material of the machine stock in the stock leaving the blending point and determining a flow target for the flow of one or more stocks arriving at the blending point and/or a flow target for the flow of the stock leaving the blending point. The es- sential idea further comprises adjusting the flow and/or consistency of one or more stocks arriving at the blending point and/or the concentration of the sec- ond raw material of the machine stock in the stock arriving at the blending point in such a manner that the flow of one or more stocks arriving at the blending point follows the determined flow target and/or the consistency fol- lows the determined consistency target and/or the concentration of the second raw material of the machine stock in the stock follows the determined target concentration and/or adjusting the flow and consistency of the stock leaving the blending point and/or the concentration of the second raw material of the machine stock in the stock leaving the blending point in such a manner that the flow of the stock leaving the blending point follows the determined flow target and the consistency follows the determined consistency target and/or the con- centration of the second raw material of the machine stock in the stock follows the determined target concentration. According to the essential idea of the in- vention, a predicted consistency change can be used instead of a consistency prediction, and a predicted concentration change can be used instead of a prediction for the concentration of the second raw material of the machine stock in the stock. In a preferred embodiment of the invention, a model predic- tive control method is used for controlling the operation of the stock prepara- tion, comprising a process model descriptive of the process or a part thereof and optimization in such a manner that the cost function associated with the optimization is minimized for optimal control of the operation of the stock preparation. According to a second preferred embodiment of the invention, dy- namic process models are used as process models.

[0013] An advantage of the invention is that the stock preparation is rapidly and exactly able to respond in different states of paper machine pro-

duction changes, such as paper web breaks, paper machine start-up, grade changes and speed changes. The solution presented eliminates the presently very common vibrations in stock preparation flows, surface levels, consisten- cies and concentrations and, consequently, the effect of these disturbances on paper quality, and enables a much more accurate adjustment in time than methods being used at present.

[0014] The solution of the invention is quite similar in the production of board and soft tissue, and, consequently, in the present description, the term'paper'refers not only to paper, but also to board and soft tissue.

[0015] The invention will be described in detail in the attached draw- ings, in which Figure 1 schematically shows a stock preparation department in a paper machine, Figure 2 schematically shows the operational principle of controlling a machine stock flow, Figure 3 schematically shows the principle of determining the total amount of the flow of component stocks arriving at a blending/machine chest, Figure 4 schematically shows the principle of controlling component stock dosing from a component stock chest, Figure 5 schematically shows the principle of diluting a component stock after the component stock chest, Figure 6 schematically shows the principle of dosing a component stock from a stock tower, Figure 7 schematically shows the dilution of a component stock after a stock tower, Figure 8 schematically shows the calculation of a component stock consistency prediction, Figure 9 schematically shows the determination of a chest output flow without a dilution step, Figure 10 schematically shows the modelling of a dilution step, Figure 11 schematically shows dosing into a blending chest, Figures 12 and 13 schematically show the modelling of a flow into a chest, and Figure 14 schematically shows the principle of the adjustment solu- tion used in the solution of the invention.

[0016] Figure 1 schematically shows a stock preparation depart-

ment or stock production and dosing line in a paper machine. Figure 1 shows the paper machine 8 very schematically way by means of a rectangular box. In the stock preparation of Figure 1, machine stock KM to be fed to the paper machine 8 is composed of three component stocks OM1, OM2 and OM3, which are mixed with each other. For the sake of clarity, the dosing line of only the first component stock OM1 is shown in its entirety. The dosing lines of the second component stock OM2 and the third component stock OM3 are sub- stantially similar. The dosing line for component stock OM1 includes a stock tower 1 acting as the storage chest for component stock OM1. From the stock tower 1, component stock OM1 is fed with a first pump P1 along a feeding pipe 2 to a component stock chest 3 acting as a dosing chest. From the component stock chest 3, component stock OM1 is fed with a second pump P2 along a dosing pipe 4 to a main line 6 in the stock preparation, leading to a blend- ing/machine chest 5, to which main line 6 components stocks OM2 and OM3 are led in the same way. The component stocks OM1, OM2 and OM3 start to blend with each other in the main line 6, but more efficient blending of the component stocks OM1, OM2 and OM3 occurs only in the blending/machine chest 5, where efficient blenders are used to blend the component stocks OM1, OM2 and OM3 with each other. From the blending/machine chest 5, the machine stock KM composed of the component stocks OM1, OM2 and OM3 is fed with a third pump P3 along a machine stock dosing pipe 7 to the short cir- culation of the paper machine 8 and further to the headbox for feeding the pa- per stock to the wire section of the paper machine 8. The stock preparation of Figure 1 includes three component stocks to be blended with each other, but it is evident that the number of component stocks used in the production of a paper web may vary such that one or more component stocks are used in the production of the web. Typically, 2 to 6 component stocks are used in the pro- duction. Furthermore, Figure 1 shows the blending chest and the machine chest as a combined blending/machine chest 5, but they may be, and usually are, physically entirely separate chests.

[0017] The consistency of the paper stock fed into the wire section of a paper machine typically varies between 0.3 and 1.5%. At an upper section 1a of the stock tower 1, whereto new component stock OM1 is fed, the consis- tency of component stock OM1 is typically 10 to 14%. Thus, component stock OM1 has to be diluted before being pumped to the paper machine 8. The component stocks OM1, OM2 and OM3 are diluted by addition of dilution water

into the stock in such a manner that the consistency of the machine stock KM to be fed in due course into the short circulation is about 3%. As dilution water is typically used white water, which is separated from the short circulation of the paper machine 8 and from which fibers and fine matter and ash are usually removed with a disc filter. The component stocks are diluted in several steps.

Figure 1 shows the dilution of component stock OM1 with dilution water fed immediately after the stock tower 1 at a blending point DP6 to the suction side of the first pump P1 via an adjusting valve V6 and a dilution water duct DW6.

At this point, the consistency of the stock is diluted from a consistency level of 10 to 14% to a level of 5 to 6%. After the component stock chest 3, component stock OM1 is further diluted with dilution water fed at a blending point DP4 to the suction side of the second pump P2 via an adjusting valve V4 and a dilu- tion water duct DW4, typically to a level of 3.2 to 3.5%. The component stock dosing line may comprise a plurality of successive component stock chests and, after them, blending points, but for the sake of clarity Figure 1 only shows one component stock chest 3. One more stock dilution step is usually arranged between the physically separate blending and machine chests. Component stock OM1 can also be diluted in a lower section 1 b of the stock tower 1 by re- cycling the stock and adding dilution water to component stock OM1 at a blending point DP7 via an adjusting valve V7 and a dilution water duct DW7.

[0018] The size of the stock tower 1, the component stock chest 3 and the blending/machine chest 5 depends on the production capacity of the paper machine 8 and the paper qualities produced with it, wherefore the size of the chests may vary significantly. When the same paper grade is manufac- tured at all times, larger chests are used than when the paper grade changes very often. In newspaper mills, large stock towers 1 and component stock chests 3 are typically used. In this case, the volume of the stock tower may be up to thousands of cubic meters. In a fine paper mill manufacturing a plurality of paper grades, the stock tower 1 or the component stock chest 3 may have a volume of some tens of cubic meters only. The stock tower 1 is usually consid- erably larger than the component stock chest 3 and the blending/machine stock chest 5.

[0019] When the basis weight of paper is being adjusted, a basis weight adjustment unit 9 requests for the necessary fiber flow or machine stock KM flow. Since the desired grammage is generated from the fiber flow in the paper machine, the solution of the basis weight adjustment is based on equa-

tion MS-LBW=F-Cs. k, (1) wherein MS is machine speed at reeler [m/s], L is web width at reeler [m], BW is basis weight caused by fibers or dry weight of paper if no filler is metered [g/m2], F is machine stock flow KM [I/s], Cs is machine stock consistency KM [g/l] and k is adjustment factor that takes into account the stock loss in the long circulation and the portion of rejects in the short circulation.

Since the basis weight adjustment unit 9 cannot be aware of future stock con- sistency variations, the term'at 3% consistency'is added to the machine stock KM flow request, i. e. in all cases the desired fiber flow is led to the paper ma- chine. The machine stock KM is pumped from the blending/machine chest 5 with pump P3 along the machine stock dosing pipe 7 to the short circulation of the paper machine 8. From the blending chest, new stock is pumped into the machine chest in such a manner that stock flows at all times from the machine chests via an overflow bin back into the blending chest. This ensures a con- stant state for machine stock KM pumping and, similarly, the machine chest contains a constant amount of machine stock KM at all times. The blending chest surface level varies, and blending chest surface level measurement is used to adjust the flow of component stocks entering the blending chest in or- der to keep the surface of the blending chest at the desired level. Process- dynamically, the blending chest is an integrating chest, wherefore the adjust- ment of the blending chest surface level is slow and results in overshoots, since as the output flow increases for instance 0.01 m3/s, the flow into the blending chest has to be momentarily changed to level 0.02 m3/s before the surface is at the desired level. In this way, a single machine stock KM flow change increases gradually towards the stock towers 1 as high as up to 0.2 m3/s, and presently used adjusting methods are unable to adjust the basis weight BW of the web sufficiently rapidly in a controlled manner.

[0020] The solution of the invention for controlling the operation of the stock preparation in a paper machine 8 utilizes the ability of model predic-

tive control (MPC) to calculate a prediction, i. e. future control commands, for control messages required for controlling the operation of the stock prepara- tion. These calculated control predictions are utilized by shifting the flow data and prediction for component stocks OM1, OM2 and OM3, determined based on the machine stock KM flow data and prediction, which take into account the machine stock KM flow change caused by the basis weight adjustment unit 9 backwards along the process flow produced by the dosing line, whereby the adjustments guiding the process stock flow forward utilize the predicted future flow changes. This means that the prediction for the feed flow of a given chest can be used as an indication of the amount of stock to be pumped from the chest. Besides stock flows, stock dilution steps and stock consistencies also need to be administered, and therefore the consistency data and consistency prediction transferred forward along the dosing line can be used by dynamic process models to predict and take into account the stock consistency varia- tion caused by stock flow changes when adjusting the fiber flow and dilution.

[0021] In the following, the operation of the solution of the invention will be studied by way of example in the adjustment of the basis weight of a paper web. The operation of the stock preparation is dividable into sections and, for the sake of clarity, the solution of the invention is presented by means of subprocesses descriptive of the stock preparation portions.

[0022] When the basis weight of paper is being adjusted, the basis weight adjustment unit 9 requests for the required fiber flow or machine stock KM flow from the blending/machine chest 5, from where the dosing of machine stock KM is controlled with a first control unit CONTROL1. The first control unit CONTROL1 controls the dosing of machine stock KM by controlling the third pump P3 or by controlling the set value of flow control. Flow control can also take place by means of an adjusting valve mounted after pump P3. The spe- cially structured valve is called a basis weight valve and it is extremely accu- rate. Generally, flow control or flow rate control can be carried out by changing the valve opening, pump speed or rotational volume or all these manners known per se. The basis weight BW of a paper web is measured at the dry end of the paper machine 8, for example immediately before the reeler, whereby the basis weight adjustment unit 9 requests for the necessary machine stock KM flow based on the difference between the desired basis weight BW value and the measured value. Machine stock KM flow control constitutes a first subprocess 10, which is schematically shown in Figure 2, which also shows

the inner operation of the first control unit CONTROL1 as a block diagram. In the first step, a consistency prediction KMCsPr for the machine stock KM dis- charged from the blending/machine chest 5 and determined at the previous calculation cycle is read. Machine stock KM consistency DT1 is then meas- ured, and it can be either total consistency or fiber consistency, and a machine stock consistency prediction DT1 Pr is calculated based on the machine stock output consistency prediction KMCsPr and the measured machine stock con- sistency DT1. A machine stock flow FT1 is then measured and a machine stock flow control set value FIC1 and a machine stock fiber flow target value trajectory KMFFTr, i. e. volume flow, wherein machine stock consistency is 3%, calculated by the basis weight adjustment unit 9, are read. The calculated ma- chine stock consistency prediction DT1Pr, the measured machine stock flow FT1, the machine stock flow control set value FIC1 and the machine stock fiber flow target value trajectory KMFFTr are used as the basis in model predictive control, i. e. MPC, to calculate a machine stock control message KMFmv, which may be a new flow control set value FIC1 or a control message SIC1 for the speed of a corresponding actuator, in this case pump P3. The measured ma- chine stock flow FT1 and machine stock control message KMFmv are used to calculate a new machine stock flow prediction KMFPr, i. e. a prediction stating how much machine stock KM is pumped from the blending/machine chest 5 to the paper machine 8. The MPC model includes a response from the PID con- trol's set value to the flow. This is a preferred alternative, since known methods of control engineering enable the determination of a response regarding how a control circuit has to behave in set value changes. In accordance with known tuning methods of control circuits, a control circuit can be tuned to give said re- sponse. MPC evens out the flow target within the control performance limits optimizing the cost function generated by the output error and the control change. The machine stock flow prediction KMFPr is relayed further to a sec- ond control unit CONTROL2 controlling the dosing of component stocks OM1, OM2 and OM3.

[0023] Figure 2 does not show the dilution step between the blend- ing chest and the machine chest. If said step is in use, the solutions presented for component stock OM1 dosing and dilution in Figures 4 and 5 can be used.

[0024] Modern paper machine basis weight control calculates sev- eral future flow changes for the fiber flow, which constitute the future target value trajectory. Based on this information and by calculating the future consis-

tency trajectory of the preceding chest, an optimal flow trajectory can be ad- justed and it is implemented by flow control. Since this method provides, at the dilution step, information about the consistency trajectory of the stock arriving at a blending point, i. e. future consistency, and the flow trajectory and consis- tency target of the stock leaving the blending point are also known, an optimal dilution water or component stock flow trajectory can be set, which is imple- mented by flow control. Instead of stock consistency, the concentration of the second raw material of the machine stock, such as various fillers, additives or chemicals in the stock, can also be controlled, i. e. in addition to or instead of stock consistency, the solution of the invention can be used to control the con- centration of fillers, additives or chemicals in the stock. Dilution water and vari- ous fillers, additives and chemicals can also be added at the same blending point, which is usually before the pump. Fillers, additives or chemicals can also be fed into the inlet of a chest not containing dilution. For the sake of clarity, the figures do not show the addition of fillers, additives or chemicals to ma- chine stock or component stocks, or the measurement of their concentration in the stock.

[0025] Figure 14 schematically shows the principle of model predic- tive control. Model predictive control (MPC) is a method known per se in con- trol engineering. Figure 14 schematically shows a control message 12 or a control variable 12 and a variable 13 to be measured or controlled. The point in time tO is set to correspond to the present moment, at which time history data on control variable 12 and the variable 13 to be controlled are available. The solution of the invention utilizes the capability of MPC to calculate a prediction for the process output, i. e. the variable 13 to be controlled, based on the capa- bility of MPC to calculate a so-called free or unrealised response of the proc- ess by means of previous control variables, measurements, predicted distur- bance variables, i. e. the difference between the current and desired states of the process, and the process model, whereby MPC can be used to solve the process control problem by means of the available control variables, i. e. ma- nipulable variables, in such a manner that the process output variables, i. e. ad- justable or controllable variables are as close to the target value as possible at each particular point in time. The several control changes 16 calculated for the control variable 12 by means of control and optimization and to be imple- mented in future are presented as step-like changes after the present moment to. As target value for variable 13 to be controlled by means of MPC can also

be used a time-dependent variable value, i. e. target value trajectory 14, which can be arranged to start from the last measured value of variable 13 or which may be preset, as in the example of Figure 14. Figure 14 also shows predic- tions 15a and 15b calculated for the controllable variable 13. Prediction 15b corresponds to a situation that arises if no new control measures are taken.

This corresponds to the initial state in optimization. When optimization calcu- lates control changes 16, the result is a prediction 15a that is descriptive of process output and is the result of the control measures. The control imple- ments the first measure and, after control interval dt, at point in time to +dt, new control changes are calculated in the same way. In addition to process models, an essential part of model predictive control is thus optimization, wherein future process controls for the desired operation of the process are determined based on the predicted disturbance variables, the cost function descriptive of the quality or target of the control, and the limitations set on the optimization.

Accordingly, the invention utilizes the capability of the optimization cost func- tion, included in MPC technology, to give a penalty for both a process output error and a control change calculated by a controller. This enables the opera- tion of the process to be stabilized and so-called soft, i. e. slowly acting, changes to be achieved, the changes yet being well timed. Literature dealing with model predictive control is abundantly available, an example being D.

Clarke : Advances in Model-Based Predictive Control, Oxford Science Publica- tions, 1994 and R. Soeterboek: Predictive Control a Unified Approach, Pren- tice Hall, 1992.

[0026] The dynamic stock making process models, used in the solu- tion of the invention, are fully known per se. For instance Donald P. Campbell : Process Dynamics, John Wiley & Sons, Inc., 1958 describes a basic theory of creating dynamic models for physical processes. The invention presents a so- lution for coupling together general models descriptive of process dynamics and model predictive controls. The present invention utilizes the capability of dynamic models to calculate a prediction for process flows, surfaces and con- sistencies and the capability of model predictive controls to bind the prediction to the last measurement result of the process and to utilize it in control calcula- tion. In addition, model predictive control couples successive control cycles to- gether in an intelligent manner both by giving control changes a penalty and by utilizing historical data on the control changes on previous control cycles. The present invention utilizes the ability of MPC to calculate a prediction, i. e. future

control commands, for a control message. This control prediction is utilized by conveying a flow prediction caused by basis weight control or a corresponding measure, such as grade change, backwards in the process flow, whereby the controls pumping process flow forward utilize the predicted future flow changes, whereby the feed flow prediction of the chest know how much is go- ing to be pumped from the chest. Since it is desirable to control not only flows but also dilution processes and consistencies, the consistency variation pro- ceeding with the process can be predicted and taken in account by means of dynamic process models in both fiber flow controls and dilution controls. In this case, the calculation proceeds stepwise.

[0027] Figure 3 schematically shows the determination of the total amount of the flow of compound stocks OM1 to OM3 into the blend- ing/machine chest 5, forming a second subprocess 20. Figure 3 also shows a block diagram of the inner operation of a second control unit CONTROL2 con- trolling the second subprocess 20. Component stock flows FT31 n, where n is the number of component stocks, and the surface level LT2 in the blending chest are measured and a blending/machine chest 5 surface level trajectory LT2Tr is calculated. In addition, the machine stock flow prediction KMFPr, cal- culated in the first subprocess, is read. A common component stock flow target OMFTr is then calculated using MPC. When the target value trajectory OMFTr for the total flow to be fed into the blending/machine chest 5 is calculated, component stock consistency measurements DT31 n and component stock chest output consistency predictions OMCsPr1 from the previous calculation cycle are read. These are used to calculate a consistency prediction DT3Pr1 n for each component stock. Since a given amount of each component stock is desired and the intention is to maintain the desired ratio of component stocks, a flow target OMFTr1 n is calculated for each composite stock based on the previous data by means of formulas (7), (8a), (8b) and (8c), presented later.

[0028] Figure 4 shows the dosing of component stock OM1 from the component stock chest 3, forming a subprocess 30, which is controlled by a third control unit CONTROL3. Figure 4 also shows a block diagram of the inner operation of the third control unit CONTROL3, which controls the third sub- process 30. Figure 4 only shows the dosing of composite stock OM1 from the dosing chest 3, but the principle shown in Figure 4 concerns similarly all com- posite stocks OM1, OM2 and OM3, wherefore the notations of the variables lack the subscript 1 denoting component stock OM1. A component stock flow

FT3, a component stock flow control set value FIC3 and the component stock flow target OMFTr calculated by the second control unit CONTROL2 are first read. A component stock control message OMFmv, which may be a new flow control set value FIC3 or a control message SIC3 for a corresponding actuator, in this case for the speed of pump P2, is then calculated by MPC on the basis of the measured composite stock flow FT3, the component stock flow control set value FIC3 and the component stock flow target OMFTr. A component stock flow prediction FT3Pr is then calculated based on the composite stock control message OMFmv and the measured component stock flow FT3. The component stock OM1 flow prediction FT3Pr is transferred to a fourth control unit CONTROL4 controlling the dilution of component stock OM1 and to a fifth control unit CONTROL5 controlling the dosing of component stock OM1 from the stock tower 1.

[0029] Figure 5 schematically shows the dilution of component stock OM1 after the component stock chest 3, forming a fourth subprocess 40, which is controlled by the fourth control unit CONTROL4. Figure 5 also shows a block diagram of the inner operation of the fourth control unit CONTROL4 that controls the fourth subprocess 40. If several dilution steps are in use, then in each dilution step the process is the same. Figure 5 relates similarly also to component stocks OM2 and OM3. A component stock chest 3 output consis- tency prediction OMCsPr calculated in the previous cycle is first read. The component stock consistency is then measured, either total consistency or fi- ber consistency, DT3 and a component stock consistency trajectory DT3Tr is calculated, by means of which the consistency is directed to the desired target value. The DT3Tr may also be a preset consistency target that does not change as a function of time. A dilution water flow FT4 is then measured, and both a dilution water flow control set value FIC4 and the component stock flow prediction FT3Pr calculated by the third control unit CONTROL3 are read. A dilution water control message DFmv, descriptive of the position of a dilution water duct DW4 adjusting valve V4, or as in Figure 5, the flow control set value, is then calculated based on the component stock chest output consis- tency prediction OMCsPr, the component stock consistency trajectory DT3Tr, the measured dilution water flow FT4, the dilution water flow control set value FIC4 and the component stock flow prediction FT3Pr by using MPC. A dilution water flow prediction FT4Pr is calculated based on the dilution water control message DFmv, the dynamic process model and the measured dilution water

flow FT4, and transferred further to the fifth control unit CONTROL5 controlling the dosing of component stock OM1 from the stock tower 1. A consistency prediction DT3Pr is calculated in the same way and transferred to the third control unit CONTROL3 and to the process model predicting the consistency of the blending/machine chest 5 output.

[0030] Figure 6 schematically shows the dosing of component stock OM1 from the stock tower 1, which constitutes a fifth subprocess 50 that is controlled by the fifth control unit CONTROL5. Figure 6 also shows a block diagram of the inner operation of the fifth control unit CONTROL5 that controls the fifth subprocess 50. Figure 6 relates similarly also to component stocks OM2 and OM3. The surface level LT1 in the component chest 3 is first read and a surface level target trajectory LT1Tr is computed, by means of which the surface of the chest is directed to the desired level. A component stock flow FT5 is then measured and a component stock flow control set value FIC5 is read. In addition, a component stock chest 3 output flow prediction FT3Pr- FT4Pr is calculated. The component stock flow control message OMFmv, which may be a new flow control set value FIC5 or a speed control message SIC5 of a corresponding actuator, in this case pump P1, is then calculated us- ing MPC and based on the measured component stock flow FT5, the compo- nent stock flow control set value FIC5, the component stock chest 3 output flow prediction FT3Pr-FT4Pr and the component stock 3 surface level target value trajectory LT1TR. A component stock flow prediction FT5Pr is then calculated based on the component stock control message OMFmv and the measured component stock flow FT5 and transferred to a sixth control unit CONTROL6 controlling the dilution of component stock OM1 to be metered from the stock tower 1.

[0031] Figure 7 shows the dilution of component stock OM1 after the stock tower 1, constituting a sixth subprocess 60, which is controlled by the sixth control unit CONTROL6. Figure 7 also shows a block diagram of the in- ner operation of the sixth control unit CONTROL6 that controls the sixth sub- process 60. Figure 7 relates similarly also to component stocks OM2 and OM3.

A stock tower 1 output consistency prediction MTCsPr, calculated in the previ- ous cycle, is first read. The component stock consistency is then measured, ei- ther total consistency or fiber consistency, DT5 and a component stock consis- tency trajectory DT5Tr is calculated, by means of which the consistency is di- rected to the desired target value. Further, a dilution water flow FT6 is meas-

ured, and a dilution water flow control set value FIC6 and the component stock flow prediction FT5Pr determined by the fifth control unit CONTROL5 are read.

MPC is then used to calculate the dilution water control message DFmv, which is based on the calculated stock tower output consistency prediction MTCsPr, the component stock target consistency trajectory DT5Tr, the measured dilu- tion water flow FT6, the dilution water flow control set value FIC6 and the component stock flow prediction FT5Pr and in this case is illustrative of the new position of a dilution water duct DW6 adjusting valve V6 or the flow control FIC6 set value. A dilution water flow prediction FT6Pr is calculated based on the dilution water control message DFmv and the measured dilution water flow FT6. A consistency prediction DT5Pr for the consistency after the dilution step is also calculated and transferred to process models predicting the output con- sistency after the blending point. The dilution water flow prediction FT6Pr and the measured dilution water flow FT6 are transferred further to a seventh con- trol unit CONTROL7 controlling the dilution of the stock at the lower part 1 b of the stock tower 1, if such dilution is in use at the lower part 1b of the stock tower 1. The function of the seventh control unit CONTROL7 corresponds to that of the sixth control unit CONTROL6. The stock tower 1 output consistency prediction MTCsPr can be determined by taking into account the action of the flow of dilution water at the lower part 1 b of the stock tower 1 by measuring the component stock consistency in the stock tower 1 by an indirect method or by statistical methods.

[0032] Figure 8 is a block diagram of the calculation of the compo- nent stock OM1 consistency prediction. Figure 8 relates similarly also to com- ponent stocks OM2 and OM3. In a first step, the flow prediction FT5Pr for stock flowing to the component stock chest 3 and the consistency prediction DT5Pr are read. At the same time, the component stock flow FT5 and the component stock consistency DT5 are measured. These four variables are used to calculate a component stock chest 3 feed consistency prediction or fi- ber flow prediction F5CsinPr. The surface level LT1 in the component stock chest 3 is measured and a component stock chest 3 surface level prediction LT1 Pr, and the component stock flow prediction FT3Pr and the flow prediction FT4Pr for the dilution water to be added to the stock after the component stock chest 3 are read. These four variables and the component stock chest feed consistency prediction F5CsinPr are used to calculate a volume prediction VOMsPr for the component stock chest 3 and a component stock chest 3 con-

sistency prediction OMCsPr, and these two variables and the component stock flow prediction FT3Pr, the dilution water flow prediction FT4Pr and the meas- ured component stock consistency DT3 are used to calculate a component stock consistency prediction DT3CsPr at the output of the component stock chest 3.

[0033] Figure 9 schematically shows the determination of the output flow of a chest, particularly the machine chest or the last component stock chest in the dosing line without the dilution step. The consistency of the flow discharged from the chest at a measuring point can be calculated by the fol- lowing formula Cs (t) = Csto (t-tdl)-Csto (tO-tdl) + Cs (tO), (2) wherein Cs (t) is chest output consistency [g/I], Csto (t) is stock consistency in chest [g/l], tO is calculation point in time, tdl is delay caused by flow from chest to consistency meas- urement, and t is future point in time after calculation point in time tO.

Formula (2) can be used to correct the effect of errors in the process model.

The desired future flow at the blending point can be solved by the formula F (t) = FF (t)/Cs (t-td2), t, ax > t > tO, (3) wherein td2 is delay caused by flow from consistency measurement to blending point SP and F (t) is output flow [I/s] and FF (t) is the desired fiber flow.

[0034] The dilution step modelling can be schematically expressed by means of Figure 10. Chest output consistency is obtained from the following formulas Cstou (tO-td3-td4) = Cs (tO) F (tO-td4) F (tO-td4)-F2 (tO-td4) Cstou (t) = Csto (t)-Csto (tO-td3-td4) + Cstou (tO-td3-td4) and (5)

wherein Cstou (t) is chest output consistency [g/l], Cs (t) is output consistency [g/l], Csto (t) is chest consistency [g/l], F (t) is output flow [l/s], F2 (t) is dilution water flow in dilution step [I/s], tO is calculation point in time, td3 is delay caused by flow from chest to blending point, td4 is delay caused by flow from blending point to consistency measurement, and t is future point in time after calculation point in time to.

Formulas (4) to (6) serve to determine the level of chest output consistency at the start of the calculation. This level calibrates the measurements by main- taining mass balance. Output consistency is derived from the chest consis- tency prediction taking into account the predicted changes in the chest consis- tency.

[0035] Figure 11 and the following formulas (7) and (8a) to (8c) schematically show dosing into a blending chest: K1 F1(t) X(t), (8a)<BR> Cs1(t-td1) K2 Csl (t-tdl)<BR> Cs1(t-td1) K3 F3(t) X(t), (8c)<BR> Cs1(t-td1) wherein X (t) is total fiber flow determined based on feed consistencies, F (t) is the desired total flow [l/s], K1 is the desired fiber fraction of component stock OM1,

Fl (t) is the component stock OM1 flow [I/s] corresponding to fiber fraction Kl of component stock OM1, K2 is the desired fiber fraction of component stock OM2 and F2 (t) is the component stock OM2 flow [I/s] corresponding to fi- ber fraction K2 of component stock OM2, K3 is the desired fiber fraction of component stock OM3, and FI (t) is the component stock OM3 flow [I/s] corresponding to fiber fraction K3 of component stock OM3.

During dosing into the blending chest, the desired component stock flow can be determined at the same time in such a manner that both the total flow target and the desired fiber fraction target for each component stock are simultane- ously fulfilled. The total consistency target cannot, however, be fulfilled. The formulas presented do not take into account the changes between the disc fil- ter 11 input and the return flow, for example, but they are eliminated in chest surface level management, allowing one to assume that the fiber flows to and from the disc filter 11 are the same at all times.

[0036] The modelling of the flow to a chest is schematically shown in Figures 12 and 13 by means of the component stock chest 3. The flow can be modelled by formulas Fi (t) = Fo (t) + A (LTr (t)) d ()' (9) LTr (t) = f (Lsp, L (tO)), > t > tO, (10) wherein Fi (t) is input flow to chest [I/s], Fo (t) is output flow from chest [I/s], L (t) is surface level in chest [m], LTr (t) is the desired variation curve for surface level, Lsp is the desired surface level and A is chest area at level L.

Depending on the difference between the measurement and the set value, dif- ferent target functions LTr (t) can be used. If, in addition, an unknown distur- bance flow is associated with the process, its effect can be eliminated by con- trol engineering methods known per se.

[0037] The solution of the invention thus utilizes normal process operation, and all stock preparations can be adjusted in the manner of the so-

lution presented. The solution is also well suitable for managing water cycles, whereby the water amounts and flows in chests can be managed by manage- ment of chest inputs and outputs. The solution utilizes the capability of the op- timization cost function belonging to the MPC technology to give a penalty for both a process output error and a control change calculated by the controller.

This allows process operation to be stabilized and enables the achievement of so-called soft, i. e. slowly acting, but timely control measures. The solution pre- sented also enables exact monitoring of the operation of measurements, ac- tuators and adjustments, and calling the operator if the process does not oper- ate in the way predicted by the models.

[0038] Accordingly, the disc filter 11 shown in Figure 1 or another fiber recovery apparatus is associated with nearly every blending chest. A disc filter 11 requires long-fibered stock (sweetener), to which fines and filler are bonded. The assumption in the solution presented is that the stocks to and from the disc filter 11 are at equilibrium as regards both consistency and flow, i. e. sum flow and sum fiber flow are zero. If this is not the case, both flows and their consistencies can be taken into account in the calculation of the second control unit CONTROL2.

[0039] The drawings and the related specification are only intended to illustrate the inventive idea. The details of the invention may vary within the scope of the claims. It is thus clear that the solution of the invention is in con- tinuous use in the control of stock preparation, i. e. it is not intended only for use in states of change, such as for basis weight control or another grade change relating to the product being produced. Furthermore, the solution pre- sented is usable not only in the manufacture of paper and board stock but also in other chained processes, where changing and adjusting consistencies and concentrations constitute a significant factor in the process. The control units used to control stock preparation are preferably microprocessor or signal proc- essor-based data processor units, in which at least part of the required func- tions can be implemented by software. It would also be possible to use only one single control unit for controlling the operation of stock preparation and it would implement all necessary functions, but the functions are preferably dis- tributed to several separate control units. Flows, consistencies and concentra- tions can be measured using any sensors and other measuring devices known per se.