Experimentation only improves the particular item you're making at that time. There is no connection between an experiment on a wind generator today and one you make tomorrow.
Here's a basic run down:
Almost every item that can be experimented on has stats (we'll leave things like musician instruments out of this for a bit). These stats are tied to particular experimentation categories. Wind generators have two categories: Efficiency (tied to extraction rate) and Storage (tied to hopper size).
The experimental categories are displayed as percentages between 0% and 100%. The success type of the experiment ("good success", "great success", "moderate failure") determines how much improvement (or deterioration)is applied to that percentage. For example, a "great success" increases the category +7 for every point you spent; spend 5 points and get a great success, and the percentage increases by 35.
Stats on in-game items have a minimum value (at 0% experimentation) and a maximum value (at 100%). The actual value of the stat is calculated using a linear scale between these two points and the experimental percentage you reached. You take the minimum stat value and add the stat range (Max - Min) times the experimental percentage.
For example, crafting tools have a minimum effectiveness value of -15 and a maximum effectiveness of 15. If you experiment Effectiveness to 65%, the actual stat will be:
-15 + (15 - (-15) )* 0.65 = -15 + 30 * 0.65 = 4.5
The higher you experiment a category, the closer you get to the max possible on the stats associated with that category. In elite crafting professions (mainly weaponsmith and armorsmith), items have subcomponents which also influence final stats.
The limits on the "best" item you can make is determined by the hard-coded maximum at 100% experimentation, how many points you have to spend (which increases with your "artisan experimentation" skill), and resource quality.
Resource stats determine both the maximum experimentation percentage and (along with the assembly success) the starting experimentation percentage. This is really where things start to get tricky, but the rule of thumb is to look at the stats listed under the experimental category(s) in your datapad or crafting tool, and try to find resources with those stats having high values. Learning what's considered "high value" (some resources are "capped" in certain stats), and understanding the number crunching going on behind the scenes toget the start/max experimental percentages from those stats ispart of what puts the "art" into "artisan".
Kriles Ch'artoff , Chilastra server
Master Chef (retired)
Currently doing....stuff