Abstract: Fluid models are widely used to approximately optimize the service policy in complicated processing networks, where the stochasticity is replaced by arrival and service rates. These models can be cast as structured semicontinuous linear programs (SCLP) which can be solved using dedicated simplex-based solvers. Uncertainty can appear in these models when the arrival and service rates are not known exactly. In the literature, such models are addressed via robust optimization. These robust optimization models restrict the controller, determining the service policy at each point in time, to be set at the start of the planning horizon, and do not allow to change it based on revealed information about the uncertainty. We suggest a novel partially adaptive model, in which we can change the controller based on the state of the system in predefined time points, and reformulate the resulting problem as an SCLP for various uncertainty sets.