The experimental setup for heat slug propagation at zero current and field in the QUench Experiment on Long Length (QUELL) is described and re-evaluated. Eight inductively heated and four resistively heated runs are considered and analyzed with the 1-fluid GANDALF model, which assumes perfect coupling between bundle and hole helium in the two-channel cable-in-conduit conductor (CICC). Although adequate to predict the hydraulic response, the 1-fluid model exhibits an intrinsic limitation to accurately simulate the evolution of the conduit temperature at different sensor locations during heat slug propagation. In general it appears that the analysis of this type of slow transients requires a 2-fluid model, as considered in Part II of the present study. (C) 1999 Elsevier Science Ltd. All rights reserved.