I'm late into my 5th year of study. The first project I worked under as a graduate student lasted 2 years and I published 1 paper and presented at a handful of conferences. The advisor for this project is my personal/academic advisor (not the chair of my committee, since he is in a different dept). The project I am presently working on, the chair of my PhD committee is heading up (different guy). My physics department formally requires only 1 publication for graduation. My first dissertation proposal was on the first project. I rewrote a 2nd proposal entirely based on the 2nd project. Both were approved by the committee. The two projects are related in that they are MRI related and at high field strength but that's about it as the first project was about human imaging and the second has to do with MR microscopy. The chair of my committee does not want me to include this first paper in my dissertation at all. Essentially requiring me to write 3 cohesive microscopy papers for the dissertation (1 is almost done atm) .... My question is: does a dissertation need to be a cohesive story or shouldn't it just represent the many different things I have learned/contributed to in my PhD? I would like to include the first paper I published as a chapter even though it has little in common with the rest of my dissertation. I have heard of this happening to other students and they simply put them into the dissertation as various chapters even though they may not relate. It seems to me that the two topics are about imaging two different subjects (humans vs plants/cells/mice) but fall under the same umbrella of MRI at ultrahigh field.
The dissertation needs to be whatever the thesis committee needs it to be. It's rare that anyone besides them will ever read it, so theirs is the only opinion that matters.
You should work with your committee through your advisor to decide on a format for the dissertation.
If you have other work that doesn't fit in their vision for the dissertation, it's far more important and useful to your future career and the rest of the scientific community that it is published (for most fields, as a journal article), so I wouldn't worry about it too much as long as you can graduate.
This is largely field dependent. Usually a dissertation is on a single topic or a small set of interrelated topics. It might be a single document (typical in math) or it might be a collection of papers (common in health fields IIRC). It is unlikely, but not impossible, to be a set of unrelated papers; papers on unrelated topics.
But it is your advisor who is likely the only one whose opinion is important here. There may also be a committee who must decide (vote) to accept or reject your dissertation. If they don't approve your dissertation then you are unlikely to be given a degree. So, you need to talk to them. Learning this five years in isn't very professional, however, but I'll put that on the advisor, not yourself.
A dissertation is "normally" cohesive in some sense, but that is a judgement call. Work it out with the advisor.