In reality it is not possible to separate managerial and engineering knowledge without a considerable information loss. Realistic integrated conventional managerial/engineering files are rather vague and very sparse. They are not suitable for direct applications of fuzzy reasoning algorithms. A fuzzy upgrading is a retrospective application of the knowledge of engineering methods and integration of large set of specific and rather isolated detailed information items. Any upgrading is highly subjective and ad hoc in nature. The upgrading objectivity can be increased by a set of upgrading algorithms. Two of them (discriminative power evaluation and fractal analysis) are described. The fractal analysis gives a trade-off between general and specific knowledge items and is presented separately as Appendix. A case study (an upgrading of sugar-cane-plant knowledge base) is presented.