I have created a cloth simulation for a flag which is running just fine (for one flag). I baked to memory and I got a nice flag waving in the wind. Hence I thought I really want two flags beside each other, waving in the wind. So i duplicated the mesh and the wind force field and moved it to the side. This second cloth mesh does not seem to catch the wind (or only slightly so) and the baking of the second force field also only shows a lesser number of frames baked (45 instead of 450, so every 10th).
I tried all sorts of things, e.g. applying the location and scale / rotation of the second force field and second flag. Sometimes this seems to have the second flag at least catch some wind, sometimes not. And it it catched the wind, it did 75 frames later. Duplicating for a 3rd flag and repeating the application of scale and rotation and location did not have the 3rd catch the wind - I am not sure there is a link.
I also tried baking to disk; and I followed the tips I found to delete the existing bakes for rebaking by changing settings, so that the old bake is effectively removed. I know it was removed as all blue lines vanished. But the outcome is still the same. The first flag simulates and bakes like a star; its exact duplicate does not seem to work at all.
Does anyone have any idea what I am overlooking? It might be a pretty simple thing…
If required I will upload the .blend file. I am attaching a picture of how the basic set up looks like.
Thank you for looking at this, maybe someone can help me see the light and get my second flag feel the wind 
See if this works:
Add another Cloth Cache then delete the existing one (flag_2_cache). Then rebake.
Thank you Richard - this helped a lot in getting rid of the old cache and if I bake a new one, I am now getting the full 450 frames baked. But still the second flag catches no wind. I think that due to the fact that I might have copied (not by intention) the flag at frame 75 earlier (and not by frame 1) has corrupted the physics computation for the file for this mesh. I created a quick new file and duplicated a subdivided mesh and force fields and added cloth modifiers separately and run the simulation and it worked just well - both flags catch the wind.
I also tried to completely remove the cloth modifier from the second mesh and re-add a new one. This worked for the flag catching the wind, but it still catches too late and too few amounts of wind to make a proper simulation. But it does not seem to be deterministic. If I add another mesh with a new cloth modifier, this one does not catch the wind…
Maybe I just need to redo the whole file from the start. Key learning for me: do not copy any meshes with baked physics on them - copy all first (from a file which has the set-up before physics simulation is baked the first time).
EDIT: I am now pretty sure that the initial file was somehow corrupt. I took the new file and duplicated a mesh with bake. Rebaking got me problems with the bake - but wind was catching. I used Richards method to get rid of the bake and now the duplicated mesh simulates fine.
I am not marking this solved, because I still do not know what may have corrupted the initial file, but I will start working from here and update as I have more information.
RE-EDIT: I did some additional tests and I am not sure why files corrupt once you add multiple cloth simulations. I wonder if someone else has similar issues. While some files allow the smooth and easy creation of multiple clothes driven by the wind force fields, every time I wanted to gear up the complexity of the mesh to get proper sim, either the second cloth behaved funny - or it did not catch the wind anymore.
SOLUTION: Instead of creating duplicates of the meshes in separate objects, I duplicated the mesh structure within one object. This seems to simulate fine for both flags even at high complexity. I had no issues at all. Drawback: Using different materials is only possible through assigning the duplicate mesh with such a different material for those vertices. But that is easy to handle as is handling UV unwraps.
So I will mark this SOLVED for now, although the technical question on the behaviour of the separate cloth simulations remains unanswered.
