Hi Leo,
It is hard to say without knowing the details of the model. Which solver are you using? How is the quality of your mesh? It might be even due to the inappropriate temporal and/or spatial discretization.
Please refer to the online help:
"The respective fluid, heat, or mass balance is subdivided into flows related to the various boundary-condition types, contributions from distributed sources/sinks (e.g., groundwater recharge), storage capture (negative) and release (positive), and a remaining imbalance term arising from solver residual error and errors due to the spatial and temporal discretization."
(http://www.feflow.info/html/help73/feflow/11_Results/Budget_Analysis/budget_analysis.html)
If you need a further check of the model, please directly contact our support(mike.de@dhigroup.com) to send the model.
Best,
Jintao