I couldn't find any radius variable in SimpleKmeans class, however I use alternative solution.
Clustering find best cluster that has minimum Euclidean distance(or manhattan ,..), so It will outlier instance if the nearest cluster is more than a specific threshold.