Large-plot designs

Size of test site B Depends on study design. The minimum area required for a typical large-plot design is about 0.25 ha Test site must allow for test design plus sufficient buffer zone around perimeter of field to protect against external disturbance For bare-soil studies, shady sites should be avoided Continued overleaf)... 

The team has brainstormed the factors and developed a prioritized list of the key ones. For each factor, ranges were established that were well outside of the anticipated operating range in order to maximize the effect size (but not so large as to change the fundamental reaction chemistry or overly stress the process facilities). Table 5 summarizes the final list of factors and their ranges that are to be included in the first DOE. Time, of course, will also be a (nested) factor, but, as previously discussed, its levels will be determined after the whole plot design is established. 

The current practice is to rely on minicomputers with more efficient programs and faster mass storage devices to speed up the data handling and processing. (There is not much we can do about data acquisition.) Since, if the 2-D experiment is performed in any detail, the computation (and plotting) time could take many hours even with a modern minicomputer, we should rethink about how the experiments should be performed. Basically, we are back to the pre-FFT situation when it was unthinkable to do an on-line FT experiment. We believe that it makes much sense to process 2-D NMR data offline on large computers designed for ultra fast computations and let the minicomputer concentrate on data acquisition. 

The designs that we have previously discussed are based on the complete randomization principle. However, in many situations, it is impossible to randomize all treatment combinations. In such cases, the split plot design may be used. The name split plot comes from the agricultural experiment in which the whole plots are considered for a large plot of land and the sub-plots are used to represent a small plot of land within the large area. 

If sufficient data are available, much more information can be provided when different curves for various percentages of failure are plotted. Where such data are available, reasonable design criteria would be based on some probability for failure, depending on how critical the effects of failure occur. If a large, expensive repair of a complex mechanism would result from the fatigue failure of one product, then a 10 or even 1 % probability of failure would be a more likely design criterion than the 50% suggested above. 

The selectivity for various rotational speeds should be determined with stirrers of small and large Ds/D,-, while maintaining the other design and operating variables constant (see Table 5.4-26). Plots of yields of unwanted products S versus N, x99, and PA r should then be made for both stirrers to determine the independent parameter which best correlates the data for both stirrer systems. 

Untreated (control) soil is collected to determine the presence of substances that may interfere with the measurement of target analytes. Control soil is also necessary for analytical recovery determinations made using laboratory-fortified samples. Thus, basic field study design divides the test area into one or more treated plots and an untreated control plot. Unlike the treated plots, the untreated control is typically not replicated but must be sufficiently large to provide soil for characterization, analytical method validation, and quality control. To prevent spray drift on to the control area and other potential forms of contamination, the control area is positioned > 15 m away and upwind of the treated plot, relative to prevailing wind patterns. 

In reference 20, a typical robustness test is not performed, but a study on the influence of peak measurement parameters is reported on the outcome. The study is special in the sense that no physicochemical parameter in the experimental runs is changed, but only data measurement and treatment-related parameters. These parameters can largely affect the reported results, as shown earlier, and in that sense they do influence the robusmess of the method. The different parameters (see above) were first screened in a two-level D-optimal design (9 factors in 10 experiments). The most important were then examined in a face-centered CCD, and conclusions were drawn from the response surfaces plots. 

While methods validation and accuracy testing considerations presented here have been frequently discussed in the literature, they have been included here to emphasize their importance in the design of a total quality control protocol. The Youden two sample quality control scheme has been adapted for continuous analytical performance surveillance. Methods for graphical display of systematic and random error patterns have been presented with simulated performance data. Daily examination of the T, D, and Q quality control plots may be used to assess analytical performance. Once identified, patterns in the quality control plots can be used to assist in the diagnosis of a problem. Patterns of behavior in the systematic error contribution are more frequent and easy to diagnose. However, pattern complications in both error domains are observed and simultaneous events in both T and D plots can help to isolate the problems. Point-by-point comparisons of T and D plots should be made daily (immediately after the data are generated). Early detection of abnormal behavior reduces the possibility that large numbers of samples will require reanalysis. 

When there is no replication and the design is sufficiently large either three normal plots can be constructed, one for the design variable contrasts, one for the environmental variable contrasts, and one for the... 

---