...
Another proposed design was to use a series of orifices within a tube that supplies the alum stock to the untreated water. Given the head loss that these orifices would provide, we needed to determine if the resulting alum flow would be slow enough to dose the water appropriately. To do this, we allowed a 20-percent error (since the dose was not intended for flocculation) and looked at the range of flow given this allowable error and the constant decrease in alum solution level. Unfortunately, we found that the resulting flow of alum from this design was too large for our purposes.
...