One of your columns contains non-numeric information. This might be because your data frame (tdt) has a column specifying the treatment. Here is an example with the iris data set which has "Species" in column 5.
pca <- prcomp(iris) # Fails with "Error in colMeans(x, na.rm = TRUE) : 'x' must be numeric"
pca <- prcomp(iris[1:4]) # Success
Hope this helps.