A large but difficult to quantify part of the work is figuring out what questions to ask in the first place. It’s difficult to quantify because it tends to be done in “aha” moments-- You’ll get great ideas at lunch (especially if you’re eating with colleagues), or as you’re falling asleep, or on the toilet, or whatever.
Then you need to set up the problem in such a way that it’s calculable. This often involves drawing diagrams on a whiteboard, carefully labeling all of the measurable quantities on it, and recording the mathematical relationships between the quantities. Massage those relationships enough, and you end up with the equations you need to solve to answer your question.
Now, sometimes, the equations you get will be just plain unsolvable: There simply isn’t enough information. At this point you need to figure out how to get the additional information you need. Sometimes this means setting up or designing a new experiment, sometimes it means heuristic guesses or approximations, and sometimes it just means that you publish what you’ve got, and let others try to find more information.
Even if there’s enough information for the equations to be in principle solvable, sometimes the calculations will be too hard for the calculation to be remotely practical (even with computers). In this case, you need to figure out what approximations you can do (usually this involves concluding that some effects are negligible compared to others, and leaving those out of the calculations). How much you can get away with here depends on how much precision you want: Sometimes, just getting an order of magnitude is good enough, and sometimes, other calculations have already been done and your whole goal is to get them more precise than others have, by considering some of the things they neglected.
Then, once you’ve got the problem set up mathematically in this way, you put it in a computer, let it run over your lunch hour, or overnight, or over the weekend, or whatever, and get your results. Or, if the problem is big enough that just putting it in the computer will take a while, you tell one of your grad students to do it.
Of course, once you have the results, you’re still not done “physics-ing”. You also need to share those results. This, too, has many steps, ranging from telling your lunch partners about the interesting results that seem to have cropped up, to telling the other members of your group at your weekly meeting, to preparing a Powerpoint to show at a seminar to the department, to making a poster for a conference, to publishing the paper (not all of the steps always occur, of course). And there’s also some overlap between the “physicsing” and the “filling out expense reports”, since the expense reports are usually for grants, and you get grants by pointing out the interesting work you’ve done, and telling the grant agency about all the great ideas you have for new work, as soon as you have the funding.