As a PhD Candidate I was not given a manual; I was given actually written and accepted papers in my field. Different fields can have different customs; match your writing to other published papers, in the last decade or so, actually in your field.
You can find these on Google Scholar, you can search for some papers on whatever topic you want; pick out the ones free to access, and see how they are laid out; the sections and subsections. ARXIV papers are often works in progress but free to access. Find the commonalities for your field, they can be different for mathematics, computer science, geology, chemistry, etc.
You can also pick up the terminology they use, the acronyms and abbreviations everybody knowledgeable in the field knows. Pay attention to the tone, academic papers are almost never written in the first person (like "I did this, I noticed that," etc.) They are written in a neutral tone (e.g. "notice that in the domain x \in [1,2), (x-1) approximates log_2(x)..."). Or in the plural, even with a single author, much like an instructor: "Next, we want to focus our attention on the residuals after applying this transformation..."
Google and Wikipedia are your friends, here, for parsing any terminology you don't get.
Most academic journals accept PDF papers; and one of the most used tools for writing them is a language called LaTeX (pronounced "lah-tech" or "lay-tech", I prefer the first pronunciation; the X is not pronounced "eks"). That link contains an introduction to the language.
LaTex is a document presentation system, and a hill to climb all on its own; it took me many months to get even marginally proficient in it. But there is plenty of online help. It is also freeware available in many places. In my view, nothing beats LaTeX in the presentation of mathematical formulas; I've tried systems from Microsoft and others, they are garbage. I wouldn't use anything but LaTeX if you have formulas, proofs, lemmas, etc.
You write a plain text file in LaTeX (with extension .tex), and then "compile" it into a document; typically a PDF. If you made mistakes, you edit the text file and recompile it. You can write your paper in multiple files, and include them in a main file; that will let you rearrange \sections if you like. Like a line "\input s_motivations" will include s_motivations.tex in that spot. Professionally when we write papers we tend to organize them like this, even if a section, like the introduction, is less than a page long.
But nobody cares; you don't submit your .tex files; you only submit the final PDF. As the author, only you (or your collaborators) change the .tex files. Breaking the paper up into separate files per section (or even subsections) minimizes edit collisions; I can tell you I am working on the Conclusions section, you tell me you are addings details to the Testing section, so we aren't changing the same thing. Even if you are using GIT or some other repository system, avoiding edit collisions is a good thing.
The particular LaTeX "compiler" I prefer (free) is pdflatex, it produces a PDF.
Any kinds of images, including graphs, can be included in the LaTeX file with some commands that include a PDF image; these can also resize the image to fit properly in the column or on the page.
Getting images to appear exactly where you want is somewhat difficult in LaTeX, it reflows text and equations as it sees fit. But you will notice in academic papers that is accepted in academia, you may have to hunt backwards or forwards to find an image the paper is talking about; it may even be at the end of the paper.
To create graphs and images, there are dozens of tools. I personally just use the Open Office Suite, I use Excel to create graphs, then copy the graph image to Draw, and I may enhance it there, if I need to draw arrows or lines on the graph, or even add text (like a little "x" on the maximum and "m" on the minimum) that Excel cannot handle. Then I export just that image to a PDF. In Latex, the command to import the image allows for cropping and resizing of the image. I may go through a few rounds of this (messing with the image, and seeing how it looks after compile) before I am satisfied with it.
LaTeX itself (at the website I gave you) provides free help in getting started, and StackExchange right here can often answer LaTeX questions.
Academic journal sites usually have style preferences for their papers; and you can download their style files (a feature of LaTeX) that set margins, reference types, fonts, page-numbering commands, etc.
I haven't looked, but LaTeX is pretty old and the de facto standard for academic papers in the sciences; I imagine there exist plenty of free tutorials on getting started, and plenty of help for the more complex nuances you might need.
Good luck. Examine the format of existing papers in your field. From my first day as a graduate student in CS, my advisor had me doing exactly this, one or two papers a week, for months, to read and summarize and see how they are done.
LaTeX is not arcane; the commands are fairly simple and descriptive English. \label{some_text} to create a reference label; then \ref{some_text} to create a reference to that in text. So if you label a graph, you put \label{graph_1} on it, and then refer to "Figure \ref{graph_1}" in your text.
Or, to include a graphic, something like:
\includegraphics[width=2.9in, height=2in]{Method_1_Speedups.pdf}
All stuff like that. Good luck.
