Model predictive control history

In order to extrapolate the laboratory results to the field and to make semiquantitative predictions, an in-house computer model was used. Chemical reaction rate constants were derived by matching the data from the Controlled Mixing History Furnace to the model predictions. The devolatilization phase was not modeled since volatile matter release and subsequent combustion occurs very rapidly and would not significantly impact the accuracy of the mathematical model predictions. The "overall" solid conversion efficiency at a given residence time was obtained by adding both the simulated char combustion efficiency and the average pyrolysis efficiency (found in the primary stage of the CMHF). 

Controlled strain is the preferred mode of operation for nonlinear studies. In step-strain experiments, an important source of experimental error is the deviation of the actual strain history from a perfect step. Laun [96] and Venerus and Kahvand [43] have discussed this problem and how it can be addressed. Gevgilili and Kalyon [ 100] found that the actual strain pattern generated by a popular coimnerdal rheometer in response to a command for a step was, in fact, a rather complex function of time. One approach that is of use in comparing data from any transient test with the predictions of a model is to record the actual, non-ideal, strain history and use this same history to calculate the model predictions. 

In traditional process control, models are often used to predict the deviation of the controlled variable from the desired state, the process error. This assumes that one knows the desired state. In complex batch processes, the desired state of the process is also dependent on history and changing dynamically. Further, most process models have to predict the outcome of an entire cycle to determine if the product will be good, so predictions are not available in real time, even for a slow process like the autoclave cure however, partial models have been used as virtual sensors to expand on the information available from sensors [38]. Saliba et al. used a kinetic model to predict the degree of cure as a function of time and temperature in a mold and used that predicted degree of cure to time pressure application and determine the completion of cure. Others [39] have used the predictions of models together with the measured progress of the process to predict future trends and even project process outcomes. 

Fortunately, controlling separations is not nearly as complicated as much of the literature may make it seem. My aim is to cut through much of the detail and theory to make this a usable technique for you. The separation models I present are those that have proven useful to me in predicting separations. I make no claim for their accuracy, except that they work. There are many excellent texts on the market, in the technical literature, and on the Internet, continuously updated and revised, that present the history and the current theory of chromatography separations. 

Computer control of the operating parameters to maintain setpoints, the use of off-line kinetic models to determine operating conditions, and the use of online models to make predictions of the intermediate parameters and the final product properties as a function of the operating history have all become commonplace in industry over the last 30 years. This certainly has had a strong influence on the improvements in product quality and uniformity and the large increases in production rate that have occurred over this period of time. 

Using a computer-aided model for the prediction of pseu-doallergic reactions from prospective data collected from 581 patients in a controlled clinical trial with an outdated formulation of polygeline, accurate prediction of 86% of the patients who had a systemic reaction was possible (9). The data were handled by multivariate analysis using the independence Bayes model. The predictive accuracy of other reactions was poor. A history of allergy was recorded in 26% of the patients who had systemic reactions and in 12 and 13% of the patients with no systemic or skin reactions. However, these differences were not statistically significant. 

When no replicates are available, common weights that use the observed data include 1/Y or 1/Y2. These two weighting schemes in essence assume that the variance model is proportional to the mean or mean squared, respectively, and then crudely use the observation itself to estimate the mean. Although using observed data has a tremendous history behind it, using observed data as weights is problematic in that observed data are measured with error. A better estimate might be 1/Y or 1 /Y2 where the predicted values are used instead. In this manner, any measurement error or random variability in the data are controlled. 

The experimental values of the three responses for this optimized resin are presented in Table 8, together with the values predicted by the empirical models (equations (3)-(5)). The particleboard properties (IB strength and FE) are sufficiently close to the predicted values, considering the inevitable variability induced by the use of industrial grade reagents, complex control of synthesis conditions (namely the pH history and monitoring of the viscosity in the condensation step) and the natural heterogeneity of the wood mix used for particleboards. 

Experience has shown that a stress-strain relationship may fail to predict correctly the response when small reversals in the strain history occur this is the case when the structural model is subjected to a ground motion record. Initially this problem was identified in the model of Menegotto and Pinto (1973) by Filippou et al. (1983), who pointed out that, in order to avoid such an undesirable behaviour, the memory of the analytical model should extent over all previous branches of the stress-strain history. In terms of implementation this would be impractical and thus Filippou et al. (1983) proposed to limit the memory of the model to four controlling curves, which warrant that, at least at the structural level, this numerical problem is almost fully eliminated. 

To determine the extent of bloating or expansion in an industrial rotary kiln, one must carry out laboratory tests using bench scale furnaces for the evolution kinetics and further correlation tests in a pilot rotary kiln for appropriate temperature profiles. The temporal events determined are, in turn, used to plan quarry operations for product quality control. The same data may also be useful in developing a mechanistic mathematical model that can predict temperature distribution and density changes in the raw material as they journey through the kiln (Boateng et al., 1997). Such tools have proven to be useful for the control of product quality as new mines are explored or even as different strata of the existing mine are explored for feedstock. Some of these time-temperature histories are discussed herein. 

To develop an understanding of the sub-T relaxational processes of PSF and the nature of molecular motions involved, new forced torsional dynamic medianical data for PSF samples with well-controlled thermal histories were studied. To assist in the assignment of molecular motions, geometry optimized CNDO/2 (Complete Neglect of Differential Overlap) and molecular orbital (MO) calculations of model compounds were used to predict energy barriers to rotation. These energy barriers are compared to the activation energies determined from the dynamic mechanical data for each relaxation. Details of the CNDO/2 and molecular mechanics techniques used may be found elsewhere. 

---