I agree with Cristoforo's comment above, in that the source of the data (people, facility, their previous publications) is most important. I would also ask though, why are those experimental data unpublished? And how did you obtain them, are they from a colleague? If it is because the data are brand new and you anticipate that it will be published soon, then by all means use it to verify your numerical work, as it will most likely appear as published data soon anyway! If the data is relatively old and no one else has published any similar results, then I would be quite cautious. If someone has unique data, that you cannot replicate anywhere else, then why would it be left unpublished? Just some thoughts of mine that may help you.

Publishing a research work does not make it reliable, what matters is the quality of what is done. If you are sure of the accuracy of a result from an unpublished experimental work, you can use it to validate any other numerical result and this does not mean your work can't be published thereafter.

Thank you very much for all answers. However, I have been asked to review a numerical study at which the numerical results was validated against experimental tests which are unpublished in any journal or conference. The author(s) reply to me that these experimental tests were conducted by same author(s) and the experimental study is submitted to another journal and under review. Do you think that the experimental results could be used to validate numerical model even when they are under review? Thank you in advance.

A very important question, I totally agree with Dr. Olumoyewa Dotun Atoyebi