The design of a spiral inductor deals with first what semiconductor process you have to work with. The PDK or Process Design Kit should describe the process. This would allow you to determine the metal and dielectric layers you have at your disposal. Also, it should tell you what structures are proven in that semiconductor process. However, it is not uncommon for custom structures to be designed. Next, you need to decide on the criticality of L and Q for a given footprint area. Inductors are the largest devices on-chip and often one wants high Q for a given L to acheive a lower power dissipation. In some cases, lower Q is OK when used for degeneration in a circuit.
Following all of these choices, often today to achieve high Q, spirals are designed with multiple layers (up to 3) and are symetical in design with parallel paths to leverage mutural coupling and thus more L and high Q for a given area. More than 4 layers will likely not result in an efficent structure.
In order, determine what geometry is ideal you will need an advanced EM solver (and not the free one from university) to compare and make trade-offs which can include what shape, ground shield, etc... Depending on your frequency, you will need an EM solver than can model well skin effects, dispacement currents, ground effects, and handle complex shapes, especially cross-overs.