This level of system identification involves the correlation of experimental results with results from a detailed theoretical model. Such a model is generally used prior to system operation for studying system stability, for studying performance during normal maneuvers such as load changes, and for designing control systems. Also, this model used for studies of normal system operations usually includes the same modeling assumptions and system parameters used in the accident-studies model. These models have a major influence on system design and operating policies. Clearly, experi­mental verification of the model is in the best interest of safe and efficient system operation.

Differences between experimental results and theoretical predictions may be due to errors in the model, errors in the parameters in a correct model, or both. The identification of parameters in a correct model can be approached in a systematic way, as described in the next section. The adjustment of theoretical models is not so systematic, but a good under­standing of the underlying processes that are responsible for the calculated behavior can aid in this. For example, Fig. 6.4 shows theoretical and experi­mental phase shift results for the molten-salt reactor experiment (53). The

predicted bump in the neighborhood of 0.3 rad/sec did not appear in the experimental results. Prior theoretical work had shown that this bump was due to pure time delays associated with external-loop circulation. This knowledge permitted an adjustment of the model by allowing for more mixing in the external-flow circuits.