My best guess would be that the K values are calculated using the minor loss coefficient based on experimentally determined head loss. (but don't quote me)
You might be able to get K values from a given manufacturer or supplier of the parts you're using. Without that, you can probably get pretty close using some assumptions.
One way to do it would be to just treat the small bends as 45 bends, which would give you a conservative estimate (since you'd be overestimating friction in each smaller bend). I don't expect that the difference would be too huge (Though it depends on how many there are, and how long the run is).
Another option to get a little closer would be to look at something like this table, http://www.metropumps.com/ResourcesFrictionLossData.pdf which provides loss coefficients for mitre bends.
Regarding the reducers, typically a supplier will have a K value, IF NOT, there are some formulas which can get you these values, though you'd need some other information about flow regime/pipe friction (reynolds #, friction factor)
Hope this helps!