Accuracy of neutroninelastic process

I’m trying to validate my geant4 simulations of neutroninelastic capture in Li-6 against experimental data but getting quite discrepant results for the number of detected neutrons (factors of ~3 in some cases). In particular I’m using geant4.10.7 with neutronHP to simulate a moderated Cf-252 source decaying, and a detector material enriched in Li-6. I then specifically tally the number of neutroninelastic events stemming from Li-6. I also simulate nearby lab geometry to account for additional scattered neutrons.

Are there any good resources showing the expected accuracy or existing validation studies for a similar case, or even documentation on the accuracy of the cross section in this regime?