Short answer: Between 1.5 and 2.0 for most structures.
Long answer: it depends on the applicable codes. ASD tends to use a single "safety factor" per calculation, although multiple load combinations may be investigated. LRFD/LFD design approaches tend to use multiple "safety factors", some tied to the type of loading, some tied to the material or failure mode. The combination of those factors results in a similar "net safety factor", but not identical.
Given the breadth of the question, I'd say that it's somewhere from 1 to 10. If you move outside of building structures, you'll have an even higher range. On the lower end, you have factors of safety against probabilistically determined failure level loads (Earthquake, some types of wind evaluation). On the high end you have things like overhead suspension using cables and similar things. I think you could even make an argument that the effective factor of safety for seismic is actually less than one, given that you're generally assuming that the allowable limit state is significantly less than what would otherwise be an acceptable limit state, or if you start pulling in the importance factor.
This is incredibly code, material, application, and definition of factor of safety dependent.
Factor of safety is also a pretty fuzzy term without discussing what constitutes a design load.
If you are talking about reliability indices, usually most codes try to achieve a target reliability index somewhere between 3-4 as a measure of achieving an acceptable corresponding level of safety (EDIT - at the ULS).
One standard that deals with a means of establishing a suitable reliability index from considering all possible material variations, tolerance variations, etc is ISO 2394. This standard is what code writers use to establish what strength reduction factors and load factors need to be to achieve the required level of safety and acceptable level of risk against bad stuff happening.
I'm assuming you are not talking about load factors here. That is a different question.
The safety factor is not for load and there is no dublication. On the past we was perforimg the design by WSD or ASD, so there is no factors and take a factor of safety for steel about Fy/2 and and for concerte aroun 3-5. Now a days we are doing design by LRFD for steel and limit state design in concerete due to alot of research and development. For any code IBC, ASCE, BS, EC,..etc. the same principal, there are a load factors and reduction factor as per the american codes and partial safety factors as per BS. Noting that the value of the live load itslef should be mtach with load factors and floors reduction factors from the same design code and also the resistance reduction factor with the same codes and the construction on site should follow the same code this should be match with the structure reliability.
One problem is that the "factor of safety" is not clearly defined in the first place. You'd like to think of this as actual strength/required strength or stress at failure/actual stress or something along that line. One problem is that many of the loads are probabilistic and so the design load that likely won't be exceeded is a fair margin over the expected actual load. And then if you change that probability up or down, it changes the design load, but the actual load that is seen by the item never changes in all of that.
I recall from a seminar a while back that the ASCE 7 seismic loads do not lead to uniformly-likely designs across the country- generally, the design loads are more likely in high-seismic areas than in low-seismic areas.
Safety factors cover several individual items. You seem to be primarily interested in the safety factor on the load itself. That is best described as the load factor inherent for each load type in an LRFD/LSD type load combination. For LRFD/LSD, the resistance factor (0.65 - 0.95 typically) covers things like design errors, fabrication errors, erection errors, material variations, etc.) For ASD design that all gets blended into the basic safety factor and is not distinguished between the root causes. Overall in LRFD, the net load factor generally averages out around 1.5. It is lower for loads such as dead load that you estimate with a high degree of accuracy and higher for loads (live load, snow load, wind load, etc.) that a guess at best for the proper value. Things like cranes, explosions, and other odd-balls have their own sets of factors again based on the expected knowledge of the load and its intensity.
