Abstract
We present an open, scripted reconstruction pipeline for the ADRON-55 ionization calorimeter at the Tien Shan station (3340 m a.s.l.) and apply it to 2.9×105 events from 2021 and 2023. It assembles overlapping chamber sections, applies data-driven integrity and good-run selection (16.3% flagged), and fits three-dimensional directions from hadron-block planes with material-based weights. Plane occupancy falls with depth to 0.75 (X) and 0.61 (Y) at the deepest plane of each projection, while per-channel deposition is nearly uniform across X planes, revealing a projection-dependent efficiency offset whose size, and for the deepest plane pair whose sign, depend on the amplitude threshold. One-pitch lateral profiles resolve a narrower gamma-block component, RMS 0.638±0.016 against 0.727±0.007 m; the gamma figure is an upper limit, since the events carrying half its amplitude clip the digitizer, and without them it is 0.507 m. For 990 single tracks with three active planes in each projection, the azimuthal flatness statistic improves from χ2/ndf≈1.3×103 to 15.8 after assembly, seeding, and alignment. We then locate the residual: in four fifths of that sample a plane is lit edge to edge and carries no core, so its centroid reports the aperture rather than the particle, and the alignment maps those events onto a zero-slope line that pins their azimuth to the detector axes. Removing them leaves χ2/ndf=2.0, and 1.2 on events held out of a re-derived alignment; a detector-level simulation carrying the measured channel efficiencies then reproduces that sub-sample with an isotropic cos3.5θ parent. The code and diagnostics are released, making the figures reproducible from raw data.