Four of safety injection (SI) nozzles in a 1000 MWe-class Optimized Power Reactor (OPR-1000) are fitted with thermal sleeves (T/S) to alleviate thermal fatigue. Thermal sleeves in #3 and #4 of Younggwang (YGN) & Ulchin (UCN) nuclear power plant are manufactured out of Inconel-600 and fitted solidly without any problem, whereas YGN and UCN in #5 and #6, also fitted with thermal sleeves made of Inconel-690 for increased corrosion resistance, experienced a loosening of thermal sleeves except T/S of YGN #5-1A. To identify the root cause of T/S loosening, three suspected causes were analyzed: (1) the shear force of flow on the T/S when the safety SI nozzle was in operation, (2) the differences between Inconel-600 and Inconel-690 in terms of physical and chemical properties (notably the thermal expansion coefficient), and (3) the positioning error in explosive expansion of the T/S as well as the asymmetric expansion of T/S. It was confirmed that none of the three suspected causes could be considered as the root cause. However, after reviewing the design change history from the Palo Verde nuclear power plant to YGN and UCN #3,4 and #5,6, it was realized that the two stage design modifications (in terms of groove depth & material) made an additional explosive energy required by 172% in aggregate, but the amount of gunpowder and the explosive expansion method were the same as before, resulting in insufficient explosive force that led to poor thermal sleeve expansion. T/S measurement data and rubbing copies also support this conclusion. And the T/S loosening was also attributable to lenient quality control before and after fitting the T/S that resulted in significant uncertainty. In addition, it is our judgment that the acceptance criterion applicable to T/S fitting was not strict enough, failing to single out thermal sleeves that were not expanded sufficiently. Lastly, the acceptance criterion was scientifically revised to incorporate the thinning effect and inherent uncertainties of measurements.