The institute I am studying at has added the following requirement recently:

A PhD scholar must publish two papers in his/her corresponding field
in SCI journals.

This means that a student cannot get PhD without two SCI rated papers. To the best of my knowledge, novelty is a mandate for such journals; hence, novelty has became mandatory for a student to get PhD in my institute.

In this context, I have the following query: Why is novelty mandatory for a PhD degree?

Consider a scenario in which a student is ambitious and wants to do research on an open problem or famous unsolved problem in his/her field. The student cannot attempt to research such a topic since research may be locally saturated, and he/she cannot come up with a novel idea. Does the rule not restrict the research on such problems? I can see the alternatives like tutorial papers, survey papers, etc., but some of these require only highly experienced geeks.

Note: Assume that neither the PhD student nor the PhD supervisor is an extraordinary person in that particular field.

Novelty is a basic requirement in research because without novel results, the work is of at best limited usefulness. It's not useless - review papers are useful - but it's less interesting to the people at the frontier. In the case of a PhD, it is supposed to signify a student is now capable of doing research. Research (as opposed to literature review) necessarily involves new and novel results. Therefore novelty is a requirement.

If I understand your last paragraph right, you're wondering about open problems which might be too complex to solve in one go. But in that case, you still get new and novel results - they might be partial results, but they're still results. The same goes for research using new methods that turn out not to work: knowing where not to look is not as useful as knowing where to look, but by narrowing the parameter space, it's still useful.

This answer will be a bit different and not limited to the sciences as the other current answers seem to be. It will also apply to many similar degrees not designated PhD. I'll use the term "doctorate" as an abbreviation for "research doctorate" to distinguish it from other degrees such as those that are clinically focused, for example.

Study for a doctorate (PhD and similar degrees) in a field is intended to teach you how to carry out scholarship in that field, whatever it is. Scholarship is always intended to be the extension of knowledge. A successful scholar extends what is known in the field of interest. In literature, for example, you can study Dr. Faustus by Goethe and give a perspective never considered before. This, as are many (most?) studies very narrow. But it can also be very deep. If someone says that your thoughts aren't actually novel, but a simple extension of things written a few (or a lot of) years ago, then those thoughts don't really extend knowledge. So, if all you can accomplish are derivative works, without novelty, then you haven't reached the point of being able to extend knowledge in your field.

The actual accomplishment need not be earth shattering in its implications. Most are not - even in mathematics and the sciences. But you have to be able to say something that is interesting to other scholars, preferably something that will permit them to extend their own work in new ways not yet considered.

There are a lot of ways to demonstrate that you are, indeed, able to extend knowledge in your field. A requirement of publication in a good journal is just one way and it is easy for the faculty to judge, since the reviewers and editors of the journal help them in their judgement task. Not so many years ago, in mathematics, there was no requirement to publish before obtaining the doctorate and your "proof of ability" was measured only in your dissertation as evaluated by the local faculty, perhaps with some supplementing.

Research in any field is directed at extending the "known world". You need to demonstrate the ability to do that to earn a doctorate and to be known as a "doctor" of the field. Only "novelty" will do.

Axiom

My answer derive from following definition of the PhD and its thesis.

The purpose of a PhD is to train a research student as an autonomous scientist
and a good researcher — i.e., as someone deserving the grade of Doctor.

The objective of a PhD thesis is hence to demonstrate to one’s peers that its author can be considered in this way by reporting the successful* completion of a high-quality piece of research.

Depending on one's field, the monograph-style thesis can be replaced by a set of published peer-reviewed paper. The objective remains the same.

Question

The question is then "What is an autonomous scientist and a good researcher?"

Being a good researcher means first to do science well, second to do good
science, and third to do a lot of science.

Doing science well means following its ethical rules, being thorough, honestly and appropriately reporting one’s results, crediting one’s peers for their work and acknowledging one’s work limitations, using the adequate tools and methods to solve a problem, and doing research that is reproducible.

Doing good science means addressing a relevant scientific gap, an issue that is influential, a research problem that matters for science, the society, and the industry.

Doing a lot of science means to tackle a large-scale problem, to address subsequent research gaps, to contribute to multiple issues of a coherent sub-field.

One must, however, note that these criteria are conditional to one another. Doing a lot of science serves no purpose if one does not do good science. And more importantly, doing good science serves no purpose if one does not do science well.

To this, I would add that (it is my opinion, and one may disagree) the job of a researcher is to do science.

And doing science means reporting new knowledge that is as truthful as possible.

This is achievable using the scientific method that is commonly acknowledged as the best method for creating/unveiling (depending on your ontological positioning) reliable new knowledge.

Answer

So, to pragmatically answer your question:

• 
  the purpose a a PhD is to train you as a scientist,


• 
  producing new (and highly reliable) knowledge is quintessentially the role of a scientist.

Hence, you cannot be recognized by your peers as a scientist (i.e., being awarded the _Doctor_ title) if you don't prove that you can produce new reliable knowledge by yourself.

The confusion might come from the fact that — as it has been highlighted in comments — scientists do literature review, and literature review are useful and valuable. However, they are not the core of what is a scientist.

Post-scriptum

It it worth highlighting again the last quoted paragraph:

These criteria are conditional to one another. Doing a lot of science serves no purpose if one does not do good science. And more importantly, doing good science serves no purpose if one does not do science well.

Thus, I think that, when doing a PhD, one should mainly focus of the quality of one's contribution, rather than breadth or scale.

In other words: what matters is not that much what you have proven, but how you have proven it.

Of course I had the dream of a breakthrough thesis.

Yet I think that, during a PhD, it is more important to produce only a few robust, well designed experiment and well reported results on a minor "research gap"; rather than aiming to tackle a large scale issue without being able to produce robust results or reliable conclusions.

It's only once you'll have proven that you are a good researcher and autonomous scientist, that you'll be able to secure a job as a scientist/researcher.

And then you'll have time (and money, and support, and experience, …) to shoot for the stars.

Yet this is not the objective of a PhD. Everything in its own time.

_{Citations are excerpt of… the preface of my PhD thesis.}

_{* Note that "successful" is to be understood as "We have completed each step of the scientific method". I.e., it does not consider whether results are positive or negative.}

A PhD thesis demonstrates the candidate's ability to perform research. Research, by definition, adds something novel to the body of knowledge.

Research: The systematic investigation into and study of materials and sources
in order to establish facts and reach new conclusions
(Oxford English Dictionary)