The full- and reduced-order fault detection filter design is examined for fault diagnosis in linear time-invariant (LTI) systems in the presence of noise and disturbances. The fault detection filter design problem is formulated as an H problem using a linear fractional transformation (LFT) framework and the solution is based on the bounded real lemma (BRL). Necessary and sufficient conditions for the existence of the fault detection filter are presented in the form of linear matrix inequalities (LMIs) resulting in a convex problem for the full-order filter design and a rank-constrained nonconvex problem for the reduced-order filter design. By minimizing the sensitivity of the filter residuals to noise and disturbances, the fault detection objective is fulfilled. A reference model can be incorporated in the design in order to shape the desired performance of the fault detection filter. The proposed fault detection and isolation (FDI) framework is applied to detect instrumentation and sensor faults in fluid transmission and pipeline systems. To this end, a lumped parameter framework for modeling infinite-dimensional fluid transient systems is utilized and a low-order model is obtained to pursue the instrumentation fault diagnosis objective. Full- and reduced-order filters are designed for sensor FDI. Simulations are conducted to assess the effectiveness of the proposed fault detection approach.