Re: microstrip lyout
When a wave travelling down a line hits a discontinuity, some of that power is distributed to higher "evanescent" modes of propagation. The power typically can not stay propagating in these higher modes, so it recombines with the main power a little further down the line. If you modelled this effect (by looking far away from the discontinuity), it would look a lot like a lumped element reactance had occured at the discontinuity. "Far away" means a couple of wavelengths away.
Sometimes, but not always, the physical shape correlates to the lumped element discontinuity model. A change of width in microstrip, for example looks like a shunt capacitor on the wider side (excess capacitance from the fringing fields on the sharp corners) and a series inductor (as the currents get "squeezed" into the smaller width). Sometimes the model has weird things like negative capacitances, just because they curve fit the data best.
You need to use these models if you are operating at a high enough frequency for them to have an effect, or if you have a frequency-critical function (like a narrowband filter).
If you have a number of discontinuities that are "close" together, the lumped element models break down. You then have to do a full Emag field analysis of the structure.