Then it sounds as if we are getting more shrinkage in the cement combination paste that in the cement paste.
If the materials are less absorptive, the matrix formed in the combined paste must be allowing the formation of some form of structured shrinkage with the bigger porositiy.
This points again to first a formation of the bigger matrix, the additions upon reactivity taking part, and gaining strength, and then shrinkage happening supported upon such bigger matrix.
If such is the case, ensuring such hardened matrix can't form beforehand would be the way to go.
If coming from over-reaction, the additions being too fine could be the cause, but the formation of structures before the proper time could point exactly the contrary way, too coarse grains promptly reacting at their surface and allowing the bridging.
In any case, for mixes of materials well known to the chemical experts, I would look at the granulometry of the materials (and purity of the water).