Interesting question. I am not sure what approach you used to translate a customer service level to a safety stock, nor whether you did a static or dynamic safety stock. However, I don’t think those matter so much to the question.
The real question is how would you go about simulating fluctuations in “real” demand and how (well) they are satisfied.
For example, you could create a “replenishment plan” (a series of Sched Recpts) to satisfy your projected demand plus SS. Set your planning time fence beyond your supplies, or even as "do not generate" planned orders. Then, change your demand and look for the result.
Alternatively, you could limit the supply with a constraint and again look at results with different demand patterns. (Maybe your constraint available would be your initial weekly supply plan?).
Depending upon the time horizon of the above, they could easily be overly pessimistic. Unless the supply chain has hard constraints, it is usually a question of how long it takes to react to a change in demand rather than not being able to change your supply plan at all. To simulate this, I would only look at changing demands within cumulative lead times. Have a series of SRs out to lead time for all your parts.
I believe it was the planning time fence that was the big issue. Given that we want to look out at least a year to model seasonality, the planning time fence would need to be at least a year out, right?
It is really the issue of capacity that I would like to address. Often inventory is managed on a min/max basis, SS typically being the min. Other ways of doing this is to use a range of customer service levels to determine the min and max inventory levels. The min/max can also be expressed as periods of cover.
I was looking to use RapidResponse as an analysis tool to set the inventory policy that would require minimum inventory to obtain maximum customer service using historical demand patterns, including seasonality.