High shares of renewable energy sources require transmission expansion planning (TEP) to account for flexibility resources that substitute for lines. This paper quantifies the cost of planning those resources separately from the network. National and cross-border transmission, battery energy storage systems (BESS) and hydrogen (H2) subsystems are co-optimised across 15 pathways and 12 scenarios on an 84-node European network, with demand-side management (DSM) at three participation levels and a fixed generation fleet. Storage and transmission substitute for one another in 45 of 48 paired comparisons, and in every one where national lines can be built: BESS reduce annualised network investment by a median of 3.5–7.8% and grid capacity needs by up to 12.4%, and hydrogen by at most 1.3%. The transmission scope reverses the curtailment effect of hydrogen: the H2 turbine reduces curtailment in all 24 comparisons restricted to cross-border lines and increases it in 22 of 24 comparisons that admit national lines. Physical and economic criteria rank storage technologies in opposite orders. Pathways without BESS require 5.4 times more grid capacity per unit of storage, whereas hydrogen returns 62–73% against 13–17% for BESS, excluding avoided scarcity costs; on the same basis, cross-border lines return 31% and national lines 56%. Planning storage and transmission separately costs 0.18–1.67% of annual system cost. DSM improves adequacy and defers capital without lowering operating cost: investment falls by 5.8–10.7% while the cost of building and operating the system rises by 1.2–4.6%. National reinforcement, not cross-border lines, serves all demand; without it, 5–24 TWh/year are unserved.
Networked battery energy storage systems (BESSs) require accurate state-of-charge (SoC) balancing under cell aging and explicit safety enforcement at the power-sharing layer. This paper proposes an adaptive, degradation-aware, and safety-constrained power-allocation framework for networked BESSs, built on three integrated layers: (i) a nonlinear adaptive observer that estimates SoC, ohmic resistance R0, and usable capacity from local measurements, aided by a zero-sum probing signal whose amplitude follows a closed-form design rule linking measurement noise to a target identification accuracy; (ii) a degradation-aware allocation law that shifts power burden away from high-resistance cells through a tunable resistance penalty; and (iii) a safety layer that computes the exact Euclidean projection of the raw commands onto the feasible set defined by SoC, voltage, current, and cell-power limits, tracking the aggregate demand when feasible and derating it otherwise. Uniform ultimate boundedness of the observer error is established with an explicit Lyapunov matrix, and the estimation–safety cascade is shown to preserve aggregate feasibility, with respect to the limits evaluated on the estimated states, to machine precision independently of estimation quality. Simulations on 4-unit and 16-unit networked BESSs show that the scheme reduces the R0 estimation error from 15.2% to 2.0%, lowers the high-R0 loss fraction by up to 8.3% under non-uniform aging, and achieves 5.6% lower total ohmic loss than a consensus-based SoC balancer. Multi-seed robustness studies confirm graceful degradation under measurement noise, observer-side parameter mismatch, and communication delays and packet loss.
