|
The ever increasing demands for higher production rates, paper quality, sheet uniformity, and machine efficiencies has lead to an increased demand for machine optimization. One of the tools used to optimize the wet end of the paper machine is the stock approach diagnostic evaluation more commonly referred to as a pulsation analysis.
The purpose of a stock approach diagnostic evaluation is to define those variables contributing to the non-uniform delivery of stock to the wire. Uniform delivery of stock is essential to uniform machine direction basis weight control. The non-uniform delivery of stock onto the wire can be a result of consistency, pressure, total head, vibration, control, or speed instability.
DIAGNOSTIC FUNDAMENTALS. Machine direction basis weight variation is detrimental to product quality and often results in increased operating costs. Quality parameters such as strength, caliper/density, permeability, and printability can be adversely affected by non-uniform weight. Excessive variability can also significantly reduce paper machine runability and converting efficiency, which results in lower tonnage and increased costs/ton. In severe cases, where average weight must be increased to meet minimum standards or to maintain machine runability, fiber consumption and, thus, direct costs are increased.
A comprehensive stock approach evaluation will compare the variations found in basis weight to the variations found in the stock approach system and headbox pressure through the electronic monitoring of machine process variables. Through the introduction of pressure sensors into stock lines, triggers on stock delivery rotating elements, inline measurement of stock consistencies, placement of vibration sensors on machine structural elements, and electronic monitoring of machine process signals, such as basis weight and opacity, electronic data is simultaneously taken regarding machine performance. Simultaneous sampling of multiple signals is essential to determine the propagation of the variations through the system.
Selection of the proper sensors is also critical to the analysis process. Pressure sensors that are capable of providing DC level pressure as well as good high frequency response should be utilized. Vibration on the headbox is typically measured in acceleration, since the acceleration of the headbox, primarily in the machine direction, can cause velocity fluctuations in the jet. Obtaining a trigger pulse off a rotating element is often difficult; however, a tachometer capable of measuring in reflective or proximity mode will provide a better chance of establishing the signal.
The variations found in basis weight and headbox pressure are referred to in terms of high frequency and low frequency variations. High frequency variations are typically seen as barring on the fourdrinier table and in the sheet at periodic or randomly spaced intervals. Periodic high frequency variations are generally associated with rotating elements around the wet end of the machine, including pumps, screens, wire return rolls, etc. Cavitation from a headbox recirculation valve or stream flow valve, however, can also cause high frequency random barring.
Through the years, high frequency analysis has become a fairly standard form of analysis, and many paper mills have implemented routine maintenance programs or purchased online monitoring systems capable of monitoring high frequency variations. Although the high frequency analysis is still very important, the demands for improved product quality and reduced operating costs have placed a greater emphasis on the low frequency or long-term analysis. Low frequency variations have a significant impact on the sheet 2-sigma values, and the reduction of these variations go directly to decrease operating costs and increase machine and converting efficiencies.
Low frequency or long-term variation analysis can be periodic or random in nature and is typically measured from 0 to 1 Hz. Low frequency periodics are usually attributable to very slow rotating machine elements such as rectifier rolls or showers in air-padded headboxes or worn, damaged, or improperly seamed forming fabrics. A bad or improperly tuned controller can also cause a control loop to oscillate at long-term periodic time intervals.
Sources of random long-term variations are typically more difficult to identify than those for periodics. Sources can include poor mixing of thick stock at the fan pump, flow instability, uneven mixing and/or non-stable flow of retention aid, deaerator or silo level variations, and poor control of thick stock consistency, to name a few.
When performing a low frequency analysis, the approach system is completely instrumented to measure all parameters simultaneously, including pressure, consistency, speed variation, flow, and basis weight. Simultaneously sampling all signals and presenting the data in time strip chart format provides a comparison of the variations as they travel through the approach system. This information can then be used to visually compare one signal to another in an attempt to determine the part of the process that is contributing to the basis weight variation.
To more clearly determine the similarity and direction of propagation between signals, an analysis technique called cross correlation is used. This method provides a measurement of the similarity between any two signals and the time lag between them. The cross correlation data is presented as a plot of correlation coefficient versus time.
Besides the correlation coefficient, which indicates similarity between signals, time lag is very meaningful in cross correlation analysis. For example, in a stock approach system, pressure waves can travel in both directions. When long-term pressure fluctuations are evident, the time lag indicates the direction of propagation for the pressure wave between two points. A positive time measurement would indicate the signal considered as the stimulus occurred before the signal considered as the response. The limitations on sampling points and the resolution of the analysis make it impossible to exactly measure the point of origin, but it can often be determined that a pressure wave originates at some point between two taps.
Altering machine process variables is another analysis technique used in low frequency analysis. Measuring the system response when turning off controllers, adjusting valves, raising or lowering silo or deaerator levels, or adjusting consistency to change fan pump speed and flow rates is very useful in defining the source of the low frequency variation.
The following case histories cite specific examples of low frequency problems encountered at paper mills. The analysis techniques previously described were used to identify and resolve the machine runability problems and improve machine efficiencies. Low frequency problems are not limited to specific grades, machine types, or speeds. These examples are taken from machines that operate from 500 fpm to 5,500 fpm producing heavy board to tissue.
CASE HISTORY 1: CONTROL. Control problems can be a result of a bad controller or poor control scheme in any part of the process. Control problems can make unwanted changes to machine variables, causing the appearance of a pulsation type problem. The following case history details a rush/drag control scheme problem that was causing poor runability and excessively long grade changes. Changes to machine speed or total head would cause the machine rush/drag to swing out of control, resulting in poor machine runability and reduced operating efficiency. Based on the extreme upsets in the total head signal from the DCS system, the problem was believed to be pulsation related.
This machine involves a fourdrinier unit with top wire former and fully hydraulic headbox producing linerboard from 2,000 fpm to 2,500 fpm. A pulsation eliminator tank, three pressure screens, and a fan pump with a variable speed drive set to constant speed make up the stock approach system. The header recirculation line was piped to the suction of the fan pump with the control valve located on the machine room floor close to the header. The machine also ran with speed optimization-a control process to optimize production through various machine control parameters including speed and steam consumption.
Figure 1 is a one-hour trend taken from the DCS of the total head in rush/drag control and with rush/drag control suspended. n the first part of the graph, the actual total head signal is continuously cycling above and below the set point in an attempt to achieve the set point request. It is also apparent from the graph that the total head set point was also changing in an attempt to optimize machine performance.
Cross correlation of the pulsation signals acquired during this analysis indicated the variation started in the headbox and traveled backwards through the approach system. Based on this information the variations in the stock approach system were eliminated as a possible source, and the analysis focused on the headbox.
Further investigation of the problem indicated the machine was using headbox slice position to control the rush/drag. With this control scheme the fan pump speed was set to provide the required total head for the machine speed, and the slice position was changed in an attempt to control the rush/drag.
|
FIGURE 3. Basis weight and consistency strip chart comparison
|
One of the main problems associated with controlling rush/drag with the slice position is the backlash in the jackscrews. The backlash prevents the ability to position the slice with sufficient accuracy or repeatability. As a result, the controller opens or closes the slice beyond the exact position required and cycles back and forth trying to achieve the requested set point. Slice binding, worn seals, and speed optimization control can also compound this problem.
With the rush/drag control suspended (Figure 1) it is apparent that the controller was no longer bumping the slice position, and the total head became very stable. Based on this information the recommendation was made to change the rush/drag control scheme to set the slice opening to one position and use fan pump speed control to change the total head to achieve the desired rush/drag.
With the source of the instability identified, the mill revised the rush/drag control based on this recommendation. After the rush/drag control was modified, the grade change time was dramatically reduced, speed changes did not upset the machine operation, and machine-operating efficiencies improved dramatically. Figure 2 shows a one-hour trend from the DCS of the total head during three speed increases. The typical cycle or hunting for the correct total head has been completely eliminated, and the total head signal now follows the set point very closely.
CASE HISTORY 2: CONSISTENCY. The use of good piping principles and paying close attention to flow velocities is essential to uniform mixing of stock. This is especially true for proper mixing of thick stock at the suction of the fan pump. Improper mixing will lead to mass variations in the approach system that will transfer into the sheet as basis weight variations.
|
FIGURE 4. Basis weight comparison before and after piping modification
|
This case study involves a twin wire tissue machine with fully hydraulic headbox, running from 5,000 fpm to 5,500 fpm. The stock approach system on this machine utilizes a pre-dilution and dilution system. Thick stock is brought into the suction of the fan pump in the pre-dilution system at approximately 3% consistency and is diluted with whitewater from the silo to approximately 1%. The pre-diluted stock then enters the suction of the headbox supply pump where it is further diluted with whitewater from the same silo to approximately 0.25% before being pumped to the headbox. The pre-dilution system consists of a constant speed fan pump and pressure screen. A variable speed fan pump to supply stock to the headbox is the only rotating element in the dilution system. The headbox header recirculation line is piped to the suction of the headbox supply pump with the control and manual valves located in a horizontal run in the basement.
|
FIGURE 5. Basis weight comparison headbox shower on and off
|
This machine was experiencing excessively high machine direction basis weight instability. A comprehensive pulsation study performed on this machine indicated that the basis weight variation was not a result of pressure variations in the approach system. However, variations from a bypass type consistency meter installed in the approach piping before the headbox header indicated very good signal correlation between thin stock consistency and basis weight.
The strip chart graphs in Figure 3 present a comparison of the basis weight signal, single pointed in the center of the sheet, and thin stock consistency before the headbox header. A calibration was performed on the basis weight head with the amplitude presented in lbs/ream. The calibration process for the consistency meter is rather arduous and is normally not performed unless the customer wants to know the magnitude of consistency variation. From a troubleshooting standpoint, the only thing of importance is whether or not there is signal correlation, i.e., cause and effect.
The comparison of the two strip chart signals definitely indicates the signals are very similar. However, to determine the actual similarity between the two signals, a cross correlation analysis was performed, indicating a 0.912 correlation coefficient between the thin stock consistency before the headbox header and basis weight with a time delay of 1.8 sec. With the signal correlation established between thin stock consistency and basis weight the next objective was to determine the origin of the consistency variation so corrective action could be taken.
To determine the origin of the consistency variation, cross correlation analysis was performed between basis weight and thin stock consistency at the suction of the headbox supply pump after the pressure screen in the pre-dilution system and in the silo. The additional cross correlation data indicated that signal correlation between thin stock consistency and basis weight could only be established at the suction of the headbox supply pump. No signal correlation could be established between thin stock consistency in the pre-dilution system or silo with basis weight.
Once the origin of the consistency variation was isolated to the suction of the headbox supply pump, the problem was immediately associated with poor dilution water mixing. The piping from the pre-dilution system entered the drop leg from the silo to the headbox supply pump at a 90-degree angle approximately four feet above the centerline of the pump. It was discovered that the injection piping was cut off flush at the inside wall of the drop leg, causing the flow to be injected perpendicular to the flow from the silo. It was also discovered that the calculated flow velocity from the pre-dilution system was extremely low. The combination of low flow velocity and poor stock injection was believed to be the source of the poor mixing.
Based on this information, piping modifications to improve dilution water mixing at the suction of the headbox supply pump were designed by the mill. The piping modifications were constructed from materials onsite at a minimal cost to the mill and installed during a scheduled outage. As a result of these piping modifications, basis weight variability was greatly improved, reducing operating costs and increasing machine efficiency.
Figure 4 presents a comparison of the basis weight strip chart signals for the before and after piping modification. Both signals are presented on the same scale for ease of comparison.
CASE HISTORY 3: PERIODIC LOW FREQUENCY VARIATION. This study presents data from a machine that was experiencing a periodic registration problem at approximately 1.3 sec or 0.75 Hz. Feedback from the converting process indicated that the registration problem had the same time interval regardless of machine speed. The constant time interval eliminated all elements that were machine speed related and brought the focus of the investigation to only constant speed sources.
The problem involves a two-ply board machine with an air-padded primary and secondary headbox. The primary stock approach system consists of a constant speed fan pump, cleaners, pressure screen, and bias and trim stream flow valve arrangement. The header recirculation line is piped into the secondary cleaner accepts line at the suction of the fan pump. The header recirculation valve is located in a horizontal pipe run in the basement.
Basis weight and pressure measurements recorded on the front and backside of the headbox header and headbox pondside indicated the 1.3-sec or 0.75 Hz variation was evident in the sheet and headbox but not the headbox header. Based on this information, the analysis focused on the headbox rotating elements. Because the rectifier rolls are the last rotating elements in the approach system the initial assumption was that the periodic was coming from one of the rectifier rolls. Calculations based on the rotational frequencies of the rectifier rolls indicated the rolls were not the source of the 0.75 Hz periodic. Calculations based on the speed of the headbox shower, however, indicated the source of the 1.3-sec variation could be the third harmonic of the headbox shower rotational frequency.
To confirm the source as the headbox shower, pressure and basis weight signals were record with the shower off. Figure 5 presents a comparison of the basis weight spectra with the headbox shower on and off. This comparison conclusively identifies the shower as the source of the 1.3-second variation evident in the sheet. The periodics at one and two times the shower rotation and one times the slice rectifier roll are also eliminated with the shower turned off.
Excessive vibration and deflection of the shower, while in operation, was causing the seal to break on the side of the headbox, resulting in a total head variation at the shower rotational frequency and related harmonics. The recommendation in this case was to either stiffen the existing shower pipe or replace it with a thick wall pipe to minimize the deflection. It was also recommended to change the shower sleeves with bearings for smoother rotation.
Philip Taber is a diagnostic service engineer with AstenJohnson's Papermaking Diagnostics and Services Team based out of Charleston, S.C.
|
){kind=link}
){kind=link}