The cost for process chemicals used by the global pulp and paper industry is estimated to reach US$40 billion annually by 2010. In addition, there are escalating pressures to comply with stiffening environmental standards, coupled with the need to maintain quality output without adversely affecting the bottom line.
As a result, more cost-effective and improved process control practices are needed in the paper industry, including implementation of at-line, real-time measurement of the Kappa number and brightness of the chemical pulp.
THE PROBLEM
Kappa number (the index by which the amount of lignin in pulp is measured) and brightness play an important role in optimizing the amount of process chemicals used during the delignification and bleaching processes. They are also valuable parameters, along with pulp moisture content, of the salable finished product.
Current methods of accurately testing both Kappa number and brightness are not only time consuming, but require the use of toxic chemicals and skilled technicians. On average, a paper mill can spend as much as US$115,000 per year on consumable chemicals used for testing, plus an additional US$100,000 per year (in man hours) for sample preparation. Titration, the most common analysis technique for obtaining Kappa number, takes upward of one hour to complete a single analysis, while brightness requires a more complex sample preparation and can take 24 hours or more to complete.
Once the Kappa number and brightness values are known, the amount of process chemicals required for achieving the desired pulp brightness can be determined. Increasing the frequency of in-process measurements of Kappa number and brightness throughout the pulping and bleaching process provides the opportunity to optimize use of delignification and bleaching chemicals, and minimize unnecessary waste and pollutants.
NIR
Near-infrared (NIR) technology has proven to be a trusted and versatile analytical measurement tool in various forestry applications for many years. It is a non-destructive, non-contact materials measurement method that requires little or no sample preparation, and can analyze samples in a matter of seconds. By implementing NIR analysis, process control technicians gain the ability to collect approximately 12 times more data points in an at-line testing environment.
NIR devices use wavelengths from the range of visible light at 350 nm-2,500 nm. Throughout these regions, the light energy interacts with the C-H, N-H, O-H bonds of the sample, producing a "fingerprint" spectrum (Figure 1). This spectrum can then be correlated to the Kappa number and brightness of pulp as determined by the traditional reference chemistry techniques. It is necessary to have a calibration that mathematically relates the spectra measured by the NIR to a reference measurement such as titration. Samples that are represented by the model database will then be accurately predicted.
[FIGURE 1 OMITTED]
Spectral information in the visible region, particularly those around 457 nm, can be correlated to brightness, while the near-infrared region, which contains the chemical information on lignin residuals in pulp, can be correlated to Kappa number.
Once testing points or locations for sampling have been determined, it is necessary to build the calibration model. Model building is essentially creating an equation relating multiple variables to correlate the measured spectra with the desired unknowns such as Kappa number and brightness. This involves calculating the regression equation using the NIR spectra, Kappa number and brightness data collected from the titration and traditional brightness test data. Since reference chemistry methods are expensive and time consuming, it is important to maximize efficiency and create the best possible calibration.
The key challenges in the creation of a calibration are identification of samples to use for the model, obtaining accurate and reproducible reference assays, and the development of the calibration equation using the chemometric modeling tools that combine the spectra and reference data to form the actual calibration model.
It is paramount that the model be representative of the raw material being used. If a mill is using both hard and soft wood, then both wood types must be included in the calibration model. A model must also be validated to ensure its accuracy in predicting Kappa number and brightness. This can be accomplished while building the model simply by holding back a portion of the sample data and then using it for validation. In other words, the majority of sample data is used to build the model, which is then validated by independent testing against the unused samples.
The calibration model must be updated anytime a change occurs in the incoming feedstock, raw materials, processing conditions or instrument set-up. Additionally, periodic validation will help to ensure that the model continues to be accurate. Samples that have a higher error by reference analysis may not be well-represented, or this may be an indication that equipment maintenance is required. These samples are then used in the next version of the calibration model.
...
