A sensitivity-based approach to optimal sensor selection for process networks

The paper proposes a greedy sensor-removal algorithm and repeatedly states it determines the minimum number of sensors and, among such subsets, maximizes a Gram–Schmidt-based degree-of-observability score D. However, no proof or sufficient structural assumptions are provided. The method in fact guarantees only inclusion-wise minimality (no further single removal preserves rank), not minimum cardinality or fixed-cardinality optimality of D. The candidate solution correctly formalizes the two-stage problem, identifies the algorithm’s true guarantee, and supplies counterexamples showing failure to (i) reach minimum cardinality and (ii) achieve even 1-swap optimality of D. Key claims in the paper therefore overstate what the method achieves. See the algorithm and claims in Section 4 (Algorithm 2 and surrounding text) and the definition of D in Equations (13)–(15) for the paper’s position.