Testing the Models

The copolymer equation can be written in a general form (Equations 6-70)  

The probability arguments we have reproduced on the preceding pages can also be applied to describe the sequence of isomers in polymers like atactic polypropylene, or the placement of cis-1,4 and trans-1,4 units in diene polymers. But if we write one more probability equation you will probably hurl this book from the highest tower you can 

---

Evaluation of reactivity ratios from the copolymer composition equation requires only composition data—that is, analytical chemistry-and has been the method most widely used to evaluate rj and t2. As noted in the last section, this method assumes terminal control and seeks the best fit of the data to that model. It offers no means for testing the model and, as we shall see, is subject to enough uncertainty to make even self-consistency difficult to achieve. 

Microstructure studies, by contrast, offer both a means to evaluate the reactivity ratios and also to test the model. The capability to investigate this type of structural detail was virtually nonexistent until the advent of modern instrumentation and even now is limited to sequences of modest length. 

To test the model, first check against experimental values within the design. Second, check against rates not involved in the design. Third, predict rates and execute experiments to check the results. 

The pursuit of operations research consists of (a) the judgment phase (what are the problems ), (b) the research phase (how to solve these problems), and (c) the decision phase (how to act on the finding and eliminate the problems). These phases require the evaluation of objectives, analysis of an operation and the collection of evidence and resources to be committed to the study, the (mathematical) formulation of problems, the construction of theoretical models and selection of measures of effectiveness to test the models in practice, the making and testing of hypotheses as to how well a model represents the problem, prediction, refinement of the model, and the interpretation of results (usually as possible alternatives) with their respective values (payoff). The decision-maker generally combines the findings of the... 

Studying the effects of such variations can be most helpful in designing the next round of kinetics experiments, to test the model under conditions where its validity can be confirmed or denied. Different models can be compared to be certain that the... 

In order to estimate the extent of ozone depletion caused by a given release of CFCs, computer models of the atmosphere are employed. These models incorporate information on atmospheric motions and on the rates of over a hundred chemical and photochemical reactions. The results of measurements of the various trace species in the atmosphere are then used to test the models. Because of the complexity of atmospheric transport, the calculations were carried out initially with one-dimensional models, averaging the motions and the concentrations of chemical species over latitude and longitude, leaving only their dependency on altitude and time. More recently, two-dimensional models have been developed, in which the averaging is over longitude only. 

Computer simulations have been useful for validating a kinetic model that Is not easily tested. The model was equally capable of describing multi-site polymerizations which can undergo either first or second order deactivation. The model parameters provided reasonably accurate kinetic information about the Initial active site distribution. Simulation results were also used as aids for Interpretation of experimental data with encouraging results. 

The results from the model were compared semi-quantitatively with experimental measurements to validate the model. Because of both the extremely long time constants in the system and the variations in the ambient conditions in the laboratory where the extruder was placed, it was not possible to rigorously test the model against experimental data. To conserve demands of time and materials for experiments on the extruder, numerical experiments were used to provide data for developing an optimal control system. The goal of the numerical experiments was to develop a reduced-order model suitable for optimizing the control system. 

A number of replications under at least one set of operating conditions must be carried out to test the model adequacy (or lack of fit of the model). An estimate of the pure error variance is then calculated from ... 

The process of defining any QSPR model involves three fundamental components (i) a set of descriptors, (ii) a method to select the most appropriate descriptors, and (iii) the experimental data to train and test the model. It is important to note here that none of these components are unique and many models can be... 

The agreement between the experimental and calculated values of Ce, is excellent. The data shown in Figure 2 are for a constant bake time of 17 minutes. The upper and lower limits on define a cure window. The cure window for the low solids coating is 50 C. The model was further tested by measuring extents of reaction and temperature profiles for samples attached to different parts of a car body which passed through a pilot plant oven. This simulation tested the model under conditions where the substrate temperatures were far from constant. As shown in Table II, the agreement between the experimental and calculated values of Ce is again excellent. 

To further test the model, calculations were performed to simulate the isotopic tracer experiments presented in Figs. 9 and 10. It should be noted that while the tracer experiments were performed at 438K, the rate coefficients used in the model were chosen to fit the experiments in which chemisorbed NO was reduced at 423 K. Figures 21 and 22 illustrate the nitrogen partial pressure and surface coverage responses predicted for an experiment in which 5 0 is substituted for l NO at the same time that H2 is added to the NO flow. Similar plots are shown in Figs. 23 and 24 for an experiment in which NO is substituted for during steady-state reduction. 

Range of Variables Covered for Testing the Model of Satyanarayan, Kumar, and Kuloor (S3)... 

Later on, the authors tested the models (17) and (19) using an external data set including five molecules from the test set of Calder and Ganellin and two SKF compounds from SmithKline Beecham. With the exception of two molecules, both the models displayed good overall predictivity. Finally, they predicted seven additional compounds from the model (19) and obtained good statistics again. 

Salt spray test. The model coatings of Table I are of the high solid type used in automotive top coats. Their primary function is not corrosion protection since this is first of all a matter of phosphate layer, electrocoat and/or primer. However, the topcoats may contribute to corrosion protection by their barrier function for water, oxygen and salts. Therefore their permeability is important as one of the factors in the corrosion protection by the total coating system. We feel that a salt spray test of the model coatings directly applied to a steel surface is of little relevance for their corrosion protection performance in a real system. 

Ubrlp in the Rad6p/Ubrlp-mediated formation of a substrate-linked multi-Ub chain (31). A similar approach could be used to test the model s assumption in regard to the demonstrated Ufd4p-Rpt6p interaction. 

In order to classify promiscuous and selective compounds, we used the NB modeling protocol available in Pipeline Pilot (Scitegic) [53]. The data was split randomly into 5193 compounds for modeling and 574 compounds for testing the models. In addition to the test set, 302 known drugs were also profiled and kept separate for testing the models. All sets were checked visually to ensure that no chemical classes were overrepresented in one set or the other. 

To test the models, the training set is divided into calibration and validation sets, as shown in Table 4.26. The predictive ability of the TE and MEK SIMCA models is then evaluated using samples from all 10 classes. 

The IRT method was applied initially to the kinetics of isolated spurs. Such calculations were used to test the model and the validity of the independent pairs approximation upon which the technique is based. When applied to real radiation chemical systems, isolated spur calculations were found to predict physically unrealistic radii for the spurs, demonstrating that the concept of a distribution of isolated spurs is physically inappropriate [59]. Application of the IRT methodology to realistic electron radiation track structures has now been reported by several research groups [60-64], and the excellent agreement found between experimental data for scavenger and time-dependent yields and the predictions of IRT simulation shows that the important input parameter in determining the chemical kinetics is the initial configuration of the reactants, i.e., the use of a realistic radiation track structure. 

---