Answers are grounded in this paper and cite their sources.
District-heating dispatch models are simplified, yet their decision costs are seldom quantified. This study distinguishes estimation bias, the objective-value difference between formulations, from decision regret, defined as an L 1 -relative forward-valued schedule gap based on each schedule’s cost when re-simulated with a high-fidelity forward model incorporating exponential heat losses, computed-friction hydraulics, and consumer differential-pressure constraints. Both are assessed through a 2 × 2 loss-by-topology set of decomposition controls, in which each contrast changes exactly one phenomenon, extended by a physics-evaluation framework running from a copperplate model to station-resolved hydraulics. Aggregate losses in the copperplate enable exact separation of loss and topology effects. Results are computed for one real network and 135 factorially designed synthetic networks. The copperplate underestimates operating costs by 15.1% but produces schedules whose unprovisioned loss, valued at the marginal unit, adds 46.1%; loss-aware schedules provision this loss and remain hydraulically deliverable. Loss visibility, not spatial topology, explains the discrepancy. Losses account for 95.8% of the copperplate-to-baseline gap, while topology alone, with physics fixed, adds 0.25%. Across synthetic networks, losses explain a median 100.0% of the gap. The topology effect remains within ± 0.6 % for trunk lengths of at least 5 km and never exceeds 2.4%. Aggregate loss adders may conceal the deficiency in one network but do not transfer across scenarios. Forward evaluation of all 174 transmission stations and service laterals using real component data adds less than 1% to the fixed schedule and finds no hydraulic violation. Spatial routing is immaterial under central generation; distributed generation remains open. Results assume fixed capacities, a fixed heating curve, and radial networks.
No takes yet. Share an insight, caveat, or question.
Ruess et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: