UA cloudflare authentication

Browsing by Author "Samadi, Forooza"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Advanced Heat Flux Estimation and Non-Intrusive Flow Measurement Using Inverse Heat Conduction and Machine Learning Techniques
    (University of Alabama Libraries, 2025) Peruchi Pacheco Da Silva, Ramon; Samadi, Forooza; Woodbury, Keith A.
    Non-intrusive measurements are increasingly essential in industrial applications due to their ability to monitor processes without disrupting operations. However, most commercially available non-intrusive flow meters are prohibitively expensive. This work investigates the use of a low-cost band heater to enable both flow measurement and fault detection, offering a practical alternative for real-time diagnostics in thermal-fluid systems. The first study characterizes the transient heating behavior from a band heater to a pipe using Inverse Heat Conduction Problem (IHCP) techniques. Experiments with heating durations of 5, 10, and 20 seconds were conducted without flow, using Type-T thermocouples to record surface temperatures. Five different heat conduction models were evaluated, demonstrating that simplified models can yield comparable results to more complex formulations with significantly reduced computational effort. The system was found to reach steady thermal energization within 10-12 seconds. Notably, measured heat flux values deviated from the manufacturer's nominal specification of 58.9 kW/m2. The second study employs a coupled inverse problem to simultaneously estimate the inner and outer heat fluxes of the band heater, as well as its thermal diffusivity. A physics-based model using Green's functions and exact solutions was developed to represent the annular geometry of the heater with internal heat generation. This model successfully demonstrates the feasibility of capturing directional heat flux distributions and material thermal properties under realistic operating conditions, showing that inner outer heat fluxes.The third study presents the development of a low-cost, non-intrusive flow meter and fault detection system for monitoring steady-state water flow in stainless steel pipes. Surface temperature data collected during 60-second heating cycles were used to train machine learning models. Volumetric flow rates ranging from 5.99x10-4 to 2.39x10-3 m3/s were predicted using regression models, while classification models were applied for fault detection. The Fine Tree regression model achieved a RMSE of 1.3x10-4 m3/s and an R2 of 0.94, while the Bagged Trees classifier achieved an accuracy, precision, recall, and F1-score of 0.997. The proposed system is priced at under 10% of the cost of commercially available alternatives, making it a promising solution for cost-sensitive applications.
  • Thumbnail Image
    Item
    Heat conduction using green’s functions: partial pipe heating, a novel numerical method, and inverse heat conduction
    (University of Alabama Libraries, 2020-12) Samadi, Forooza; Woodbury, Keith K. W.; University of Alabama Tuscaloosa
    Finding the unknowns in directly inaccessible areas is one of the challenging problems in engineering applications. The fluid flow or thermal conditions of the flow inside the pipe are examples of such unknowns. Developing analytical models that provide explicit mathematical formulas is a mathematically efficient and desirable way of dealing with these challenges. This dissertation focuses on developing an analytical solution that can find the temperature distribution in radial and axial directions in the pipe wall. Partial heating along with Green’s functions is used to develop this solution. Working with Greens functions sparked the idea of using them as building blocks for a novel numerical method for solving the heat conduction equation. In the second chapter, the transient temperature response inside the pipe to the partial heating on it is found using GFs and the solution is verified using a few intrinsic verification principles. This analytical solution, then, is used in developing a non-invasive method for measuring the flow rate in pipes by measuring the temperature at a single point. Optimal experiment design methods are used to find the optimal location and time duration for performing the measurement. Chapter 3 is about developing a novel numerical solution to the linear heat conduction equation. This method uses superposition of exact solutions (SES), obtained using GFs, to evaluate the temperature and heat flux at any point of the 1-D domain. The SES method is not sensitive to the size of the grids and is much more accurate than the conventional Crank Nicolson (CN) method. This method is extended to the cases in which the thermal properties vary with temperature, later, in Chapter 4. The results confirm the grid independence of the SES and show high accuracy in its prediction when the time step is not large. One of the applications of the analytical solution obtained herein is using them in solving inverse problems. Inverse heat conduction problems (IHCPs) are used to estimate unknown heat flux functions through measuring temperatures far from active surfaces. The second part of this dissertation (Chapters 5 and 6) focuses on generalizing and optimizing two IHCP solution methods.