Partially Adaptive Robust Optimization for Uncertain Fluid Processing Networks

Abstract

This paper studies the problem of controlling fluid approximations of processing networks under uncertainty. Classical robust formulations for such models rely on static (here-and-now) control decisions that must remain feasible for all realizations of uncertainty, often leading to highly conservative solutions. We propose a partially adaptive robust optimization framework that incorporates observed system information into the control policy at the planning point, enabling less conservative and more responsive decisions while retaining worst-case guarantees. Our approach begins by modeling the fluid network dynamics as a Separated Continuous Linear Program (SCLP) in the server-effort formulation. We introduce continuous-time extensions of the budgeted uncertainty set and develop three families of uncertainty structures appropriate for fluid-flow systems. For each family, we derive the corresponding robust counterpart. Since these robust counterparts include uncertainty in the objective, they cannot be solved directly by the classical SCLP--Simplex method. To address this challenge, we develop a hybrid algorithm that integrates an extended SCLP framework with the Saddle-Point Frank--Wolfe (SP--FW) method. This hybrid approach combines tractable continuous-time optimization with efficient saddle-point updates, yielding feasible and near-optimal partially adaptive robust solutions. The resulting methodology advances robust fluid control by enabling uncertainty-aware, dynamically informed decision policies while preserving computational tractability.

Modular Bayesian Inference for Partially Misspecified Mixtures with Uncertain Class Labels
Maximilian Autenrieth, Konstantin Karchev, David A. van Dyk +2
arXiv19 Oct 2026PreprintarXiv licenceFull textCosmology StatisticsarXiv
Three-dimensional animation compression optimization based on structured processing and adaptive spatiotemporal segmentation algorithm
Xuan Wang, Xinran Yan, Biao Liu +2
PLOS One8 Sep 2026Peer-reviewedCC BY 4.0Full textZoology & animal behaviour StatisticsPLOS
Adaptive Predefined-Time Output-Feedback Control of Uncertain Nonlinear Systems with Dead-Zone Actuation via Fractional-Power Feedback
Nan Wang, Rui Meng, Fazhan Tao +1
Fractal and Fractional28 Oct 2026Peer-reviewedCC BY 4.0Analysis & PDEsCrossref
Phase Locking in Oscillator Networks with Uncertain and State-Dependent Couplings
Iva Bačić, Philipp C. Böttcher, Marc Hunkemöller +4
arXiv4 Nov 2026PreprintarXiv licenceFull textDynamical systems Optimisation & controlarXiv
Adaptive Fuzzy Fixed-Time Control for Uncertain Nonlinear Systems with Input Saturation and External Disturbances
Linlin Li, Yuhao Xing, Ying Ma
Mathematics9 Oct 2026Peer-reviewedCC BY 4.0General mathematicsCrossref

Cite

BibTeX RIS

BibTeX

@misc{bengigi2026partially,
  title = {{Partially Adaptive Robust Optimization for Uncertain Fluid Processing Networks}},
  author = {Roi Ben-Gigi and Evgeny Shindin and and Shimrit Shtern},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.25513},
  eprint = {2610.25513},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.25513},
}

RIS

TY  - UNPB
TI  - Partially Adaptive Robust Optimization for Uncertain Fluid Processing Networks
AU  - Roi Ben-Gigi
AU  - Evgeny Shindin
AU  - and Shimrit Shtern
PY  - 2026
DA  - 2026/10/18
PB  - arXiv
DO  - 10.48550/arxiv.2610.25513
UR  - https://arxiv.org/abs/2610.25513
AB  - This paper studies the problem of controlling fluid approximations of processing networks under uncertainty. Classical robust formulations for such models rely on static (here-and-now) control decisions that must remain feasible for all realizations of uncertainty, often leading to highly conservative solutions. We propose a partially adaptive robust optimization framework that incorporates observed system information into the control policy at the planning point, enabling less conservative and more responsive decisions while retaining worst-case guarantees. Our approach begins by modeling the fluid network dynamics as a Separated Continuous Linear Program (SCLP) in the server-effort formulation. We introduce continuous-time extensions of the budgeted uncertainty set and develop three families of uncertainty structures appropriate for fluid-flow systems. For each family, we derive the corresponding robust counterpart. Since these robust counterparts include uncertainty in the objective, they cannot be solved directly by the classical SCLP--Simplex method. To address this challenge, we develop a hybrid algorithm that integrates an extended SCLP framework with the Saddle-Point Frank--Wolfe (SP--FW) method. This hybrid approach combines tractable continuous-time optimization with efficient saddle-point updates, yielding feasible and near-optimal partially adaptive robust solutions. The resulting methodology advances robust fluid control by enabling uncertainty-aware, dynamically informed decision policies while preserving computational tractability.
ER  -

Abstract

Related

Cite