When the various methods of obtaining information about the reactor transfer function are compared (from the standpoint of which might be the most appropriate to use), many factors appear

1 Structure, operating conditions, and physical limita­tions of the reactor system

2 Available time, personnel, equipment, and money

3 Type of information desired

4 Accuracy desired

5 Established operating and experimental policies of the plant

The various methods described in preceding sections are treated here with this set of determining factors in mind

A distinction is made between methods involving system excitation by external apparatus and those which rely on internally generated noise. In Table 6.13 an attempt is made to evaluate the methods in a general fashion, however, it should be noted that variations in the methods or special “ground rules” for comparison may lead to exceptions

Whether an input excitation signal is applied to a neutron absorber or to a plant control device will depend primarily on the information desired and secondarily on con­venience. When this selection has been made, the signal may then be chosen to be either sinusoidal or pseudorandom In recent years there has been some preference for pseudo­random signals which simultaneously measure all fre­quencies in the band of interest Pseudorandom excitation has been preferred107 because smaller perturbing ampli­tudes can be used and less reactor time is required for obtaining a given frequency resolution (see also Sec 6-7).

For the noise methods there is no input signal injection However, there must be sufficient internally generated noise in the frequency band of interest to excite the one or more variables being investigated. If this is the case, then selection of the appropriate data-acquisition and data — processing devices is the major consideration, as is also true for excitation experiments In the following sections the types of equipment are described and their relative merits are assessed

Whether one or more than one signal is used in noise analysis depends on the information desired Transfer functions in power reactors normally require two signals Furthermore, these must have sufficient coherence, Eq 6 8, in the frequency band covered to achieve the accuracy desired, і e, the effects of the same noise source—to a larger extent than separate independent noise sources— must be seen in both signals Besides giving the amplitude and phase of transfer functions, these multisignal experi­ments also may provide insight into the cause of the intiinsic noise