Loadmodel in Gatling

I have maybe misunderstood the load model in Gatling. My app (or more concrete the web services of my native app) is supposed to deal with 35 user per seconds in peak. Having red that

is the ‘admission-rate’ as in number of user calling the services per second in i.e. a peak hour I thought I could say

setUp(scn.inject(constantUsersPerSec(10) during (5))).protocols(httpProtocol)

to simulate this in 10 seconds?
Is that correct (given that the users hypotetically uses 1 second each to complete their calls). I need to model 35 users calling the (3) web services per second and I could well enough do this for some time, but it builds up I understand.

Thank you,