SearcharxivSearch

arXiv subjects

Adam Gill

Publications and source records attributed to Adam Gill.

2 recordsLinked to original sources

Digital Twin Assessment of Filter Clogging Penalties in VFD-Driven Industrial Fan Systems

Industrial ventilation systems equipped with variable-frequency drives (VFDs) often mask the aerodynamic impact of filter clogging by automatically increasing fan speed to maintain airflow setpoints. While effective for process stability, this control strategy creates a "blind spot" in energy management, leading to unmonitored power spikes. This study applies a rapid digital twin workflow to quantify these hidden energy penalties in a standard 50 kW draw-through fan room. Using a specialized computational fluid dynamics (CFD) solver (AirSketcher), the facility was modeled under "Clean Filter" (baseline) and "Dirty Filter" (clogged) scenarios. The physics engine was first validated against wind tunnel experimental data, confirming high agreement with the theoretical inertial pressure-drop law ($\Delta P \propto U^2$). In the industrial case study, results indicate that severe clogging (modeled via a 50% effective porosity reduction) can push the fan system beyond its available pressure head or speed limits, forcing the VFD into a saturation regime. Under these conditions, effective airflow collapses by over 50% (3,806 CFM to 1,831 CFM) despite increased fan effort. The associated energy analysis predicts an annual energy penalty of 8,818 kWh ($1,058/yr). This study demonstrates how a physics-based simulation provides a defensible, ROI-driven metric for optimizing filter maintenance cycles. Keywords - industrial ventilation; digital twin; VFD optimization; filter maintenance; CFD; energy efficiency

physics.flu-dyn

Screening Canopy-Induced Wind with a Lightweight 2-D RANS Quadratic-Drag Model

Vegetation belts are widely used in urban planning to manage pedestrian wind, dust dispersion, and outdoor thermal comfort. This paper presents a compact method for predicting canopy-induced flow using a steady two-dimensional RANS model with Spalart-Allmaras closure and a quadratic drag model for vegetation. Each planting zone is defined by a single leaf area index (LAI). The workflow maps LAI to porosity using an exponential coefficient of 0.5, converts porosity to a local LAI per cell, divides by the grid-measured canopy thickness to form a leaf area density, and applies a quadratic drag term. The model is benchmarked against streamwise-mean velocity profiles reported by Liang et al. at heights 0.25h, 0.75h, and 1.25h (where h is canopy height) for a belt spanning x/H = 0 to 5. Results accurately reproduce the three canonical regions-approach, in-canopy deficit, and leeward wake, and show strong agreement with experimental data in the leeward wake, critical for sitting homes and streets near planting. The method is computationally efficient, faster than 3D CFD, uses designer-friendly inputs, and is transparent for early design screening. Implementation details, including LAI-to-porosity mapping, grid-based canopy thickness, quadratic drag, and boundary sponges, are provided to enable reproduction in AirSketcher (Polar Dynamix) or comparable solvers.

physics.flu-dyn