The difference between the 2 values for W’ is about 10 %. For this parameter, that’s to be considered ‘normal’…
That may sound ridiculous, but given the fact that there are 2 models (integral/differential), with both an average value for an unknown recovery parameter and the 1001 possible ways to deplete W’, that’s just what it is.
Some are better of with the integral model and others with the differential one. If your pattern for depleting W’ is somewhat recurring, you will have better results. If it is random, there will be more spread. Some more debate around that here:
Read the article posted a little higher and you will understand that there are conditions in which you are unable to use all W’ available.