An olive oil company has manufactured a few batches of oil in the past month, but one of them has turned out faulty

An olive oil company has manufactured a few batches of oil in the past month, but one of them has turned out faulty. This mistake was discovered too late, which the faulty oil bottles were mixed in with the authentic oils. The company wants to know if you can distinguish which bottles of olive oil are authentic and which are faulty. Explain what experiment and data analysis process would you use to do this and why? Consider advantages and limitations to your approach. 

The answer should be using PCA considering advantages and limitations. I don't really know the correct way of writing experimental design, which make me stuck right now. I urgently need this, as soon as possible. Let me know if you can help.