Experimental Data and Mathematical Models

All experimental measurements are affected by errors. In general, experimental errors are made out of systematic errors and random errors. Systematic errors show a dependence on the operating conditions and may be caused, e.g., by calibration errors of sensors. Since these errors are absent in a well-performed experimental campaign and can be corrected by an improved experimental practice, they are not considered any more in this context. 

On the other hand, random errors do not show any regular dependence on experimental conditions, since they are generated by many small and uncontrolled causes acting at the same time, and can be reduced but not completely eliminated. Thus, random errors are observed when the same measurement is repeatedly performed. In the simplest case, the universe of random errors is described by a continuous random variable e following a normal distribution with zero mean, i.e., for a univariate variable, the probability density function is given by 

This function contains the second moment or variance, a2, which measures the dispersion of the data around the mean. Repeated measurements give a sample of Ad elements of this universe the methods of inverse inference allow one to evaluate the properties of the universe from the properties of the sample [10], 

the experimental errors ey- (j = 1. AD) can be evaluated assuming the data average daw as the true value (i.e., by setting ej = dj — dm, so that the mean of errors, eav = 0, is assumed to be equal to the expected universe mean) then, the corrected sample variance. v(2 can be used as an estimate of the universe variance a2  

the term Ad — 1 represents the residual degrees of freedom of the sample after the estimation of the expected universe mean, through the computation dw. 

---

Once the best estimates of the adjustable parameters have been computed, an analysis of the results allows one to evaluate the quality of the correspondence between experimental data and mathematical model and to identify the best model among the available alternatives. This analysis consists of different steps, mainly based on... 

The experimental data and mathematical models derived for the isolated polymer chain and for the size distribution of polymer chains in the aggregate have been the basis for describing many polymer properties in rubbers and thermoplastics. Early work in this area concentrated on simplified models that were amenable to the calculation devices then available. With the advent of high-speed computers and the consequent reduction in computational labor, the more exact rotational isomeric model has been developed and fruitfully applied to many polymer problems. 

Packed column with 1-in. rings or saddles. Packing depth does not affect the relationship much (Experimental data and mathematical model.)... 

Mostafazadeh AK, Sarshar M,Jawidian S, Zarefard MR, Haghighi ZA. Separation of fructose and glucose from date syrup using resin chromatographic method experimental data and mathematical modeling. Sep PurifTechnol 2011 79(1) 72—8. 

After the first initiatives, more extensive mechanisms and consequently more realistic models were developed. The break-through came with the model put forward by Guglielmi82 in 1972. It presented the basis for various models, which have in common that they are highly empirical. A mechanism is deduced from experimental data and mathematical equations describing these data are developed. Like this relatively simple models containing several fit parameters of sometimes limited physical significance were obtained. 

The results of the experimental study and mathematical modeling of the invertase-catalyzed hydrolysis of sucrose are displayed in Fig. 8. The analysis of the output substrate concentration showed substantial substrate conversion. Therefore, the data were treated by differential equations (26) and (30), whereas the kinetic parameters were fitted using nonlinear regression. Regardless of the good agreement of the calculated and experimental values, it was concluded on the basis of a comparison of kinetic parameters obtained with those known from previous works on similar preparations of immobilized invertase [30] that this method did not provide reliable results. 

From the experimental data, a mathematical model is developed to identify significant input factors and quantify their impact on the output. From this, appropriate settings for the inputs can be determined as well as the level of control required during manufacturing. The statistical techniques to accomplish this are analysis of variance and regression analyses. 

Since the 1980s, SSITKA has been widely used to understand the formation mechanism of methane as the first paraffin in the chain. The study of the dynamics of the entire complex of reactions involved in the Fischer-Tropsch process became possible only after the development of the GC-MS technique with high resolution time. A review of field suggests that the cycle of papers by van Dijk et al. [18-21] describes the results that were obtained using the full potential of the SSITKA technique. First, a comparison of C, O, and H labeling on different Co-based catalyst formulations and in different conditions was made. For the first time, a substantial part of the product spectrum (both hydrocarbons and alcohols) was included in the isotopic transient analysis. After the qualitative interpretation of the experimental data, extensive mathematical modeling was performed for the identification and discrimination of reaction mechanisms. 

Over the years there have been many attempts to simulate the behaviour of viscoelastic materials. This has been aimed at (i) facilitating analysis of the behaviour of plastic products, (ii) assisting with extrapolation and interpolation of experimental data and (iii) reducing the need for extensive, time-consuming creep tests. The most successful of the mathematical models have been based on spring and dashpot elements to represent, respectively, the elastic and viscous responses of plastic materials. Although there are no discrete molecular structures which behave like the individual elements of the models, nevertheless... 

The studies described in the preceding two sections have identified several processes that affect the dynamic behavior of three-way catalysts. Further studies are required to identify all of the chemical and physical processes that influence the behavior of these catalysts under cycled air-fuel ratio conditions. The approaches used in future studies should include (1) direct measurement of dynamic responses, (2) mathematical analysis of experimental data, and (3) formulation and validation of mathematical models of dynamic converter operation. 

Binary solutions have been extensively studied in the last century and a whole range of different analytical models for the molar Gibbs energy of mixing have evolved in the literature. Some of these expressions are based on statistical mechanics, as we will show in Chapter 9. However, in situations where the intention is to find mathematical expressions that are easy to handle, that reproduce experimental data and that are easily incorporated in computations, polynomial expressions obviously have an advantage. 

Ideal reactors can be classified in various ways, but for our purposes the most convenient method uses the mathematical description of the reactor, as listed in Table 14.1. Each of the reactor types in Table 14.1 can be expressed in terms of integral equations, differential equations, or difference equations. Not all real reactors can fit neatly into the classification in Table 14.1, however. The accuracy and precision of the mathematical description rest not only on the character of the mixing and the heat and mass transfer coefficients in the reactor, but also on the validity and analysis of the experimental data used to model the chemical reactions involved. 

Once the initial phases have heen estimated, a model must he huilt and corrected to give the hest fit of the model to the experimental data. Refinement is an iterative process, cycling between mathematical adjustments to the protein model and examination of the model versus the experimental data and manual rebuilding. 

Our primary objective has been to present the experimental results in a convenient, combined form rather than to discuss their significance in great detail. In view of the extreme physical and chemical complexity of anthracite and the limited amount of experimental investigation to which the material has been subjected at present, an elaborate theoretical discussion would be pointless. Indeed, it is improbable that the kinetics of volatile matter release for such a complex material will ever submit to a satisfactory correlation by simple functional relationships. In spite of these difficulties, it is of interest to discuss some of the general trends exhibited by the experimental data and their interpretation by suggesting approximate theoretical and mathematical models for the release mechanism. 

To build a QSPR model, one should carefully select available experimental data, and choose the initial pool descriptors (from which the program selects the most appropriate ones) as well as a mathematical approach linking those descriptors with a given property. Then, a suitable strategy of model validation should be applied in order to obtain a quantitative assessment of the quality of predictions. Finally, some rules should be established in order to prevent the application of the models to compounds too different from those used for obtaining the models. 

---