The following issues are known to affect BlasterSim:
Some known issues related to the input file reading are listed at the end of § 1.6.
The accuracy of the sensitivities with respect to l_travel for both pneumatics and springers and vol_dead for springers may be reduced as the convergence criteria in the final time step does not consider automatic differentiation. This issue is not believed to be of practical significance, but is planned to be fixed in the future.
There may be some slight backwards motion of the projectile initially, on the order of m. This is caused by the differentiable friction model (see § 2.1.8) not having a perfect match between static friction force and pressure, and is harmless given how small it is.
When the coefficient of restitution (cor in the inputs) is zero, the sensitivity of anything with respect to the coefficient of restitution will also be zero. This sensitivity is not physically accurate and is the result of the plunger velocity being set to zero for a coefficient of restitution of zero. If a more physically realistic sensitivity is required, try a small value of cor rather than zero.
Additional issues may be available on GitHub: https://github.com/btrettel/blastersim/issues