Aiming at a kind of reentrant scheduling problem that processing time is difficult to accurately express, an scheduling model is constructed taking E/T penalty interval is as the objective. A method considering interval extreme value and interval span is designed to compare different penalty intervals for E/T quantitatively. At the same time, the proportion of interval extreme value and interval span can be adjusted according to the actual demand or the tendency of schedulers. In addition, according to the established scheduling model, a variable neighborhood chemical reaction optimization algorithm(VNCRO) is proposed. Four kinds of chemical reactions are designed. The on-wall ineffective collision adopts the idea of tracking the optimal particle in the PSO algorithm and the method of variable step size neighborhood. The inter-molecular ineffective collision makes the information exchange among the molecules in the population. Combined with the decomposition reaction and synthesis reaction, the local search and the global search reach a better balance. The reliability of the model is verified by simulation experiments. By comparing with CRO, DICA, IDABC, DWOA algorithms, the superiority of the VNCRO algorithm in resolving reentrant scheduling problem with uncertain processing time is illustrated.