Meftah Uddin

A researcher, educator, engineer and auditor. @LinkedIn and @Google Scholar
Skilled in Openstudio, EnergyPlus, Ladybug-Honeybee for Building Energy Modeling.
Experienced in Ansys Fluent for CFD and CHT.
Well versed in Python, R, SQL, Power BI for Data Analysis.

A Digital Twin Framework for Carbon Emission Monitoring and
Building Operation Feedback

This research develops a computational framework that seamlessly integrates realtime data collection, energy simulations, and performance predictions. The proposed approach is implemented through three key phases: (i) incorporating IoT sensor data into a parametric BIM model, (ii) generating training datasets using building performance simulations, and (iii) developing machine learning models for real-time predictions. The research demonstrates how this framework bridges gaps in conventional workflows, enabling dynamic adjustment of building operation and emission forecasting. This paper outlines the findings, challenges, and limitations of the framework, along with recommendations for advancing digital twin research.

Multilingual Retrieval Augmented Generation (RAG)

A multilingual RAG system is implemented using Llama Index for over 116 pdf documents.
LLM can not function factually and halucinates when domain specific questions are asked. These problems are more often seen for small models. On the other hand, in order to maintain privacy for data such as in medical, finance or manufacturing system, using commercial LLM is not recommended. RAG using opensource local LLM pipeline is an efficient and effective system to overcome these problems. We have created RAG using Llama index. Simple directory readed is used to load PDF dosuments, NLP sentencizer is used to extract sentences from documents. Sentencizer helps to create almost identical chunk from documents. The 768 dimensional paraphrase-multilingual-mpnet-base-v2 is used are embedding model and FAISS is used as Vector Store Index for quick retrival. We have used Qwen, DeepSeek, and Phi-4 mini (~3B) model as our text generation model.

Natural Convection
in Melting of PCM

The temperature and melting front profiles of PCM with the reciprocating flow arrangement are compared
to unidirectional flows in upward and downward directions. The current study examines the effects of heat transfer fluid (HTF) flow direction on the strength and duration of natural convection in PCM in a vertical cylindrical shell-and-tube container. Gallium is used as the PCM because of its low melting temperature and high thermal conductivity, and water is used as the HTF. A 2-D axisymmetric numerical model developed in ANSYS Fluent was used for the study of the cylindrical LTESS.

Digital Twin
For Smart Campus

The research proposed a digital twin framework to facilitate communication between real-time building use/operation and energy simulations to reduce greenhouse gas emissions. The paper presents the testbed development using a university campus building and the digital twin creation connected with parametric BIM, CO2 emission simulation, and ML-based prediction.

Occupancy based HVAC
and Energy Forecasting

This study addresses the challenge through occupancy-based HVAC control strategies, bolstered by machine learning predictions. The study delves into using occupancy insights for HVAC control, utilizing simulations to uncover potential energy savings. It extends its reach into time series forecasting, predicting energy patterns for short terms.

Energy optimization
in informal settlement

A massive number of stateless Rohingya refugees entered Bangladesh to escape severe crimes against humanity conducted by the Myanmar Army. Shelters made of bamboo frames, tarpaulin, and plastic sheeting creating unhealthy built environment. The goal of this research is to identify clusters of refugee shelters with the highest and lowest energy consumption rates, and the factors contributing to these differences.

Design of Experiment in
building design decision

Sensitivity analysis (SA) is important to screen out important factors in early-stage design. Design of experiment (DoE) such as fewer runs Plackett-Burman (PB) design can screen key factors comparable to thousand runs Latin Hypercube Simulation (LHS).

Solidification-Melting of Phase Change Material
for thermal management

This Electronic components generate heat during their operation, and the heat should be driven out to surroundings on a continuous basis to achieve proper functioning. Solidliquid phase change materials (PCM) have been widely examined for active thermal management of electronic devices. The advantages of PCM for thermal management are high specific heat, high latent heat of fusion, and small volume changes on phase change. PCM absorbs heat during power-on operation and releases heat at another time. In this study, we will numerically assess the effect of natural convection during melting process of PCM at vertical orientation.

Phase Change Material
In Thermal Management of Buildings

Building is one of the largest consumers of energy and is a major contributor to greenhouse gases emissions. With the global climate change toward extreme mean, the demand for HVAC system is also rising. Effective insulation with layers of different materials is considered as an efficient way for thermal insulation of a building. In this study, the insulation capability of phase change materials (PCM) is evaluated numerically and compared with wood and sand. PCM is found to be more effective to conserve the thermal condition inside conditioned space. PCM closed to the heat source is found to be better position for the thermal management.