Is a research thesis (report) with zero contribution to human knowledge acceptable?
I assume that by negative results, you mean non-significant results.
"It is an expectation that the PhD would make an original contribution and/or advance knowledge in a given field." Yes, this is true. And "X doesn't work" is a contribution to the field.
(To use an example from my area). Health practitioners are constantly dreaming up things that might work to treat various ailments (illnessess), and using them. The job of health researchers is to find out which ones work - and most of them don't work.
We used to joke that our role as health care researchers was to say no. "Nope. That doesn't work. Don't do it. No, that one's not effective either. No, don't use that. No. No. No."
Pressure for positive results just means you tweak models and data until you find them - using 'researcher degrees of freedom' (see http://pss.sagepub.com/content/22/11/1359). Your results are therefore significant, but worthless.
My PhD thesis was trying to demonstrate the nature of the relationship between stress and psoriasis symptoms (many people say "stress worsens psoriasis" - it's taken as a given truth, but it's never been empirically demonstrated). I was trying to answer things like what kind of stress, how long does it take, does it differ between people? I never found any evidence that stress did worsen psoriasis. Nor that psoriasis worsened stress (or any other psychological symptom).
A PhD is an educational process. One should demonstrate that one has learned. The most important thing about a PhD is showing what you know, what you have learned, and what you understand. If anyone gets to the end of a PhD and says "Well, those results were all positive, just as I expected", they've learned little. At the end of your PhD (or any research project) you should want to start again, and this time do it properly.
The question as currently asked is: Is a research thesis (report) with zero contribution to human knowledge acceptable?
And the answer to that question is no.
A thesis or portfolio submitted for a PhD or higher doctorate must make a novel contribution to human knowledge. It must also demonstrate that the applicant has acquired the appropriate level of research skills.
Some negative results do advance human knowledge. So a thesis with negative results and no positive results may make a novel contribution to human knowledge. e.g. demonstration of absence of an effect is a negative result, but can be a distinct and significant contribution to new knowledge (particularly if the effect was previously believed to exist).
However, just spending the time, putting in the effort, and churning out the right quantity of work, is not in and of itself sufficient.
Basis for this
This is based on a combination of my employer's guidelines, my experiences as a PhD supervisor, and advice from my colleagues. I hear that there are other (less well-respected) institutions that award doctorates just for putting in the effort and churning out the right quantity of work, regardless of novelty of contribution, or of demonstration of research skill
A quote from some official guidelines.
Here's a quote from the relevant part of the academic regulations for PhD examinations from UCL, University College London (pdf):
A thesis for the awards of EngD or PhD degree shall be examined in accordance with the criteria prescribed by UCL and the thesis shall demonstrate that it: ...
shows a student's capacity to pursue original research in the field of study based on a good understanding of the research techniques and concepts appropriate to the discipline; ...
represents a distinct and significant contribution to the subject, whether through the discovery of new knowledge, the connection of previously unrelated facts, the development of new theory, or the revision of older views;
This is a tough question. Just to abstain from the discussion of different standards in different fields, I'll talk of mathematics only. Also, I'll assume that it is a question about a PhD thesis, not about anything of lower level like Masters, etc.
The main thing is that the gap between "advancing human knowledge" and "mastering the subject" is huge and there is a lot of grades in-between. IMHO, the works that advance human knowledge are rare, be it PhD theses or papers in refereed journals. Most of us live off "doing what hasn't been done before", which is much less demanding. What I mean is that each work introducing a new idea is followed by 1000 ones applying this idea combined with already known stuff to all setups where it works. Each of those 1000 papers does what hasn't been done before but does not advance human knowledge because, once the new idea appears, every sufficiently high level professional can figure out how it may be applied elsewhere, though getting all details right may require patience and even some effort. I certainly would accept "doing what hasn't been done before" (a successful application of a well-known idea in a fairly straightforward way to a new setup) as a tolerable (but not brilliant) PhD thesis.
How much below that would I consider acceptable? Four out of every five projects I try end up in a miserable failure, when I cannot even claim that I have proved some partial result in the desired direction. I have never tried to write a detailed account of "mein kampf" for any of those (dead end moves with counterexamples at the end, chains of implications that never meet the goal, associations and studies of seemingly relevant things that failed to relate to the question at hand for some fundamental but hard to discern reason, etc.) but I would let a report like that pass as a PhD thesis if it really shows 3 years worth of high quality effort.
What I find not acceptable is a "literature survey" (understanding what is written elsewhere and relating things in a superficial way without deriving any new result or introducing any new twist into the story). In other words, my idea is that you should get your PhD after you show that you can "fight a mathematical battle on your own", not only study the battles fought by other people. The victory in a decent battle is sufficient, but not necessary. Sometimes you can be made a "general" even if you lose but show good fighting skills.
All this is my humble opinion only. As to the official point of view, in Russia we had the central committee that had to confirm every degree award before it became valid and there were written guidelines. In the USA it is way more relaxed, so 4-5 professors conspiring together can pass anyone (to the credit of them I should say that I cannot give an example of such conspiracy). Canada requires an external review to be positive (which, by the way, makes perfect sense as a simple safeguard against "local standard relaxation" to me), and so on.
As to "risky topics", the best ambitious projects are such where "something" can be done right away (not something that is worth talking about as "defendable" or "publishable", but something that shows that the student has the general grasp of the subject and decent problem-solving skills). If that something (or something equivalent) is not done within the first half a year, it is a sign of trouble and the ambitious project is better abandoned and replaced with an "apply a known idea in a straightforward way to a new setup" one. If it is, you have a chance and may consider taking the risk. Unfortunately, there is nothing that can guide you then except your gut feelings and your knowledge of the student. You and your student are in an uncharted territory all on your own, and no general advice can be given except "play by ear" and "act by circumstances".