My question is regarding applied Math journals (for example SIAM). Is there such a thing that a bad paper gets rejected very fast? Also if a paper is under review for more than 6 months, then should one expect at least a reasonable review (Although it may get rejected at the end)?
No. There are tremendous differences across journals. A "bad" paper can get rejected within hours after submission, or 6 months after submission. Good papers can also get rejected within the same timeframe. Some reviewers are very conscientious and complete the reviews quickly, but the paper can still sit with the editor for a long time. Or, more often than not, you have multiple reviewers on much different time schedules. Some journals do a very good job of triaging papers, so a rejection of a bad paper may happen quickly. But you will learn that "bad" is a very subjective viewpoint. I suggest that you do not associate the quality of the paper (i.e., good vs. bad) with the time under review.
Compared to APS journals (e.g. Phys Rev), SIAM journals do not manage their review process very well. There are no clear guidelines on how long each review step can take, and a lot depends on the attitude of a particular associate editor. On average, SIAM reviews are quite long, and the quality does not necessarily justify the time spent on it. One particular frustrating aspect is that some SIAM journals tend to offer "Reject and Resubmit" decision where a "Major Revision" would previously be offered. This is probably to improve some internal statistics (average review time, maybe).
However, it is fair to say that there are journals in applied math with even much worse review process. Ultimately, I think, this is a question of who we are as a community, and how much we want to change things. Clearly, from the editor/reviewer point of view a lazy workflow has its benefits, especially because we all have other commitments.