Center for Global Design and Manufacturing (CGDM)

The Center for Global Design and Manufacturing in the College of Engineering and Applied Science and the Smart Manufacturing Lab at UC Digital Futures is focused on research, teaching, training and industry outreach in the areas of Advanced Manufacturing and Intelligent Product Design. The Center is supported through collaborative partnerships, federal and industry sponsorships.

The Center’s research is focused on conducting research in the areas of computational tools for intelligent design and predictive modeling in additive manufacturing, manufacturing systems optimization, optimization of additive and subtractive manufacturing processes, Industrial Internet of Things (IIoT), and Augmented/Mixed/Virtual Reality in manufacturing.

The additive manufacturing research areas include multiphysics design optimization with intent for part light-weighting, predictive modelling of in-process mechanical and thermal distortion for better part quality and intelligent lattice structures. 

  • Design for Additive Manufacturing (DFAM) Constrained Topology Optimization Models
  • Multi-Material Lattice Structures for Combined Mechanical, Thermal, and Fluid Flow Applications
  • Generative Adversarial Networks for DFAM Constrained Topology Optimization Models
  • Reinforcement Learning for Generative Design of Multi-Physics Applications
  • Machine Learning Based Thermal Distortion Prediction and Compensation
  • Bayesian Networks for Process Parameter Predictions
  • Machine Learning based CAD Repair Methods for Hybrid Additive Manufacturing Processes
  • Conversational Design Assistant Based on Large Language Models such as GPT4; Applications in multifunctional multi material Topology Optimization 
  • Custom 3-D medical implants based on CT and MRI data 

The IIoT research is focused on digital twin for manufacturing (Industry 4.0 / 5.0), data acquisition and interpretation using mixed reality and computer vision for smart manufacturing.

  • Computer Vision Based Approach for Data Analytics from Legacy and Modern Machines
  • AR/VR Toolkits with Cloud Computing Based Predictive Analytics for Legacy and Modern Machines in Factory Floor
  • Mixed Reality and Virtual Reality-based Training Models for Workers; Digital Twin models for Human Centered Digital Manufacturing


Active and Recently Completed Research Grants and Contracts

  1. Intel
    • VR-based training for semiconductor chip manufacturing
  2. Siemens and Boeing (MxD)
    • UC Principal Investigator: Sam Anand
    • Smart Monitoring and Automated Real Time Visual Inspection of Sealant Application (SMART-VIStA)
  3. Digital Manufacturing Design Innovation Institute (DMDII) 
    • UC Principal Investigator: Sam Anand
    • Computer Vision-based IIoT for Smart Manufacturing of Factory Floor 
    • Other Partners: Raytheon, Faurecia, Matdan, TechSolve, ITI
  4. Industry 4.0/5.0 Institute Project
    • Cloud-based IIoT for Legacy and Smart Machines
    • VR-based Immersive Design Visualization

Computer Vision-based IIoT for Smart Manufacturing of Factory Floor 

UC CGDM, led by Prof. Anand, has experience in using computer vision algorithms in conjunction with machine learning to acquire real-time data on the manufacturing floor and convert it to neutral MTConnect streams of data that can be displayed or analyzed. Data analytics can be performed on historical manufacturing data collected. Real-time data can be collected on the manufacturing floor and assembly lines to provide instantaneous feedback for corrective actions.

  • Acquisition of data from Non-Smart equipment, Fixtures, Tooling, and Assembly process
    • Data acquisition from proprietary, legacy and non-computerized machines can be obtained from artifacts such as gauges, levers, fixtures, displays, tooling, and assembly processes from video images using image processing/machine learning/pattern classification algorithms
  • Data analytics using MTConnect® Adapters / Agents
    • The Computer Vision Toolkit controller can be seamlessly integrated with other machine data through MTConnect® applications and a real-time digital data display using ShopFloor IQ or another display software interface. The software development can be performed with open-source libraries such as OpenCV, Tesseract, Tensor Flow, and Python libraries.
  • Artificial Intelligence (AI) in computer vision
    • Computer vision and machine learning algorithms can be used for feature extraction or artifacts/parts, which can be applied directly to extract real-time data from legacy machines or systems.

University of Cincinnati collaborated with multiple industrial partners (Raytheon, Faurecia, TechSolve, ITI) to develop an innovative, affordable and non-invasive toolkit that combines currently available camera, computing, and networking technology and components with a new software analytics platform to obtain data in digital form.

Computer vision based IIoT showing process to capture legacy machines artifacts data

Computer Vision based IIoT for Smart Manufacturing on Factory Floor

Legacy machine artifact values and its ditial twin

Computer Vision for Legacy Machine Artifacts and its Digital Twin

IIoT based Augmented Reality for Factory Data Collection and Visualization

In this project, an augmented reality based IIoT application was designed at Volvo Group Trucks to collect real-time data on the shop floor machines for maintenance and troubleshooting. This included defining the data transfer mechanism to KEP Server and establishing connection between data sources and the Thingworx IIoT platform, data binding, defining spatial locations for the AR experience, and visualizing the real-time sensors status on the augmented reality app.

IIoT based Augmented reality for factory data collection in which user can visualize machine sensor locations througn an iPad

IIoT based Augmented Reality for Factory Data Collection and Visualization

Smart Monitoring and Automated Real Time Visual Inspection of Sealant Application (SMART-VIStA)

A novel smart monitoring tool using computer vision was developed as a part of MxD 20-02-07 project in collaboration with Siemens and Boeing. CGDM team at University of Cincinnati developed image processing and computer vision algorithms for detecting corner points for industrial artifacts. Additionally, a novel quantification index using shape metric was developed. The UC team also developed process parameter optimization module using Bayesian Network along with a GUI for Digital Twin.

Accurate detectino od corner points using computer vision and quality quantification using shape metrics

(a) Accurate corner points detection using computer vision (b) Quality quantification using shape metrics

Immersive Visualization of Organic Designs

Immersive Visualization enables users to leverage visualization capabilities to understand critical design and simulation parameters for analyzing computer-aided design geometries. The methodology demonstrates a virtual reality tool and interface developed through Unity-3D game use. A visualization toolkit (VTK) allows user to perform rendering of computer aided design into virtual reality environment. The visualization process allows users to represent complex structure accurately and prevent any form of potential issues. The immersive environment is evident for analyzing complex design and reduces any form of time-consuming revision and design iteration. 

Immersive visualization of designs using virtual reality

Immersive visualization of organic designs in VR

IIoT based Framework using Augmented Reality for legacy machine artifacts

The integration of legacy machines in the smart factory environment is essential for improving system performance. In this project, digitized data is obtained from legacy machines using computer vision and sent through the cloud for process automation. An AI algorithm in an AR app offers real-time process recommendations based on quality metrics input and sending data back to the machine for some of the artifacts. The AR-based Digital Thread framework visualizes machine process parameters and performance, providing real-time feedback, troubleshooting, and supporting maintenance activities.

AR-based app for legacy machine artifacts

AR-based app for intelligent predictions for legacy machines

Semiconductor Chip Manufacturing VR Training

Virtual Reality (VR) technology is being widely used for training and simulation purposes in various industries, including chip manufacturing. This study explores the potential of VR for training personnel in the manufacturing of chips from silicon wafers. The study focuses on the development of a VR-based training program that enables trainees to learn and practice the process of manufacturing chips in a simulated environment. The project includes various modules that cover the different stages of the chip manufacturing process. Trainees can interact with the simulated environment, learn the proper procedures and techniques, and practice their skills in a safe and controlled environment. 

Active and Recently Completed Research Grants and Contracts

  1. DoD-Eaton – Phase I, II
    • UC Principal Investigator: Sam Anand
    • Multiphysics Lightweight Heat Exchanger Design Optimization using Machine Learning for Metal Powder Additive Manufacturing of Lightweight Heat Exchangers
  2. Raytheon Technologies – Phase I, II, III
    • AM-based Computational Tools for Metal Powder Additive Manufacturing
  3. Digital Manufacturing Design Innovation Institute (DMDII) 
    • UC Principal Investigator: Sam Anand
    • Virtual Modeling and Certification Toolset for Metal Powder based Additive Manufacturing
    • Other Partners: GE Global Research Center, University of Illinois, Urbana Champaign, TechSolve 

Multiphysics Lightweight Heat Exchanger Design Optimization using Machine Learning

Tools such as topology optimization (TopOpt) are gaining wide recognition in the design process. Topology optimization is the computational method for finding minimum material distribution scheme for a defined loading condition within a design space without any preconceived notion for part geometry. Novel algorithms have been developed at CGDM to incorporate the DFAM constraints within TopOpt.

UC CGDM is part of a project with Eaton to generate optimized organic designs for a liquid-to-air heat exchanger suitable for the Additive Manufacturing (AM) process. The designs are modeled using the density-based multi-objective Topology Optimization (TO) algorithm. The model is constrained using Design for Additive Manufacturing (DfAM) principles and mass constraint for lightweight and cost-effective designs. A Machine Learning (ML) algorithm is used to model the turbulent airflow under dynamic loading conditions. The validation of the topology optimization model outputs (organic geometries) using functional loading and boundary conditions is performed using Siemens STAR-CCM+ simulation software.

Topology optimization for heat exchanger design

Topology optimization heat exchanger design

Geometric Feature Detection and Entrapped Powder Removal Using Computational Geometry and Graph Theory

The objective of this research is to develop a computational tool that designers can use to estimate the complexity of depowdering based on geometric features for powder-based additive manufacturing processes. The CAD geometric features are identified based on experimental testing and heuristic design principles using computational geometry algorithms. Subsequently, a graph-based algorithm calculates a range of complexity metrics representing the degree of depowdering challenges in part geometry. This research has been conducted based on experimental testing and collaboration with Raytheon Technologies. Subsequently, a powder removal algorithm has been proposed using a simulation of powder flow within the internal channels.

Powder entrapment results shown for multiple orientations

Powder entrapment results in multiple orientations

Design of variable-density structures for additive manufacturing using gyroid structure

This research presents an approach to design gyroid structures of uniformly varying density based on previously established relationship between the parametric equation of gyroids and the volume fraction. The method helps to achieve control over location based variation of the cellular density as well as unit cell size without losing any surface continuity of the gyroids. A topology optimization based methodology is used as the basis for generating variable-density gyroid lattice structures within a given design space under predefined loading conditions.

Flow for problem definition, topology optimized part, optimized variable density lattices and optimized part created using gyroid lattices

Topology Optimized Self Supporting Gyroid Lattices

RVE based density lattice structures

An optimization methodology is developed to incorporate variable volume fraction lattice structure material properties accurately (representative volume elements - RVE) into topology optimization to capture meso-structural material properties resulting in minimization of compliance for lattice geometry structural parts.

Variable density lattice structures generated using X-shaped lattice

RVE based Variable Density Lattice Structures

TopOpt Driven by Deep Learning (GAN)

Topology optimization (TO) is a powerful means of fully exploiting the geometric flexibility provided by additive manufacturing (AM). However, it often relies on iterative finite element analyses, which can entail a significant computational burden. We train a conditional generative adversarial network (cGAN) to map between the TO problem setup and optimal topology with an overhang filter. After the training, optimal topologies can be calculated in microseconds rather than minutes, with the option to post-process. Beyond this efficiency improvement, the cGAN sometimes generates qualitatively better designs than traditional TO. Additionally, generative design paradigms are applied using reinforcement learning strategies within multi-objective DFAM constrained TO models to achieve a range of organic part design outcomes for user-specified loading conditions.

Generative Adversarial Netwoek based topology optimization

GAN-based Topology Optimization

GAN-based TO model for compliance minimization

GAN-based TO model for compliance minimization trained using ~50,000 design topologies in a high-performance cluster. The TO model outputs optimal designs in seconds compared to traditional solvers that can take minutes or even hours.

Prediction of selective laser melting part quality using hybrid Bayesian network

For all the geometric flexibility additive manufacturing (AM) provides, the physics of the build process is highly complex and computationally expensive to model. We developed a Bayesian network to predict the quality of parts before they are manufactured. The network returns predictions on several part quality attributes with any available information on four key machine parameters. Predictions are highly efficient and include quantified uncertainty. Inferencing can also work the other direction (parameters from quality requirements). The chosen process for virtual demonstration is selective laser melting (SLM).

Hybrid Bayesian Network for improved part quality in additive manufacturing

Hybrid Bayesian Network for AM Part Quality

Support Parameter Tool

Support Parameters module is used to visualize the presence of support structures needed to build the part for a user-input build orientation. This tool provides quick feedback on support structures as well as calculates parameters such as support volume, support contact area, build height, and down-facing areas (areas that do not need support structures). Because this module is developed to visualize the support structures for a particular orientation, the support structures are represented as lines for faster calculations.

Support Generation Tool

Support Generation module is used to generate support structures needed to build the part for a user-input build orientation. Different types of support structures can be generated using this module, details of which will be explained in the instruction manual.

Different types of supports including solid supports, hollow supports, conformal supports and honeycomb supports.

Support Structure Generation and Visualization

Build Time Estimation Tool

Build Time is the sum of time taken by the laser to sinter each layer and the change over time between successive layers. This tool slices the body in to several layers based on the specified layer thickness. The laser path in each layer is traced and the total sintering area, sintering time across all the layers is computed. The user can specify different build direction, slice thickness, laser parameters (diameter, speed, overlap) and hatch pattern to perform analysis and identify the build parameters which result in minimum build time for a given part.

Cusp Error Estimation Tool

Cusp Error is a measure used to quantify the deviation between the designed and manufactured part. The cusp error has an effect on the surface finish of the manufactured part. This tool calculates the cusp error with respect to every facet of the CAD body and the average cusp error. All the facets satisfying the specified threshold cusp height are highlighted in green and the rest are highlighted in red. The user can specify different input parameters such as orientation of the part, layer thickness and threshold cusp height and perform analysis to determine the suitable build parameters for good surface finish.

DFAM - Small Opening Detection Tool

Small Opening Detection module is used to detect and highlight small holes and gaps in the geometry that are difficult to manufacture in AM Process. The tool can Highlight critical small opening features in the geometry that may be fused during the laser sintering process. Depends on laser diameter or nozzle diameter. This tool can rotate a part and highlight critical small opening features in the geometry and calculate the areas of the small openings in each layer.

DFAM - Thin Feature Detection Tool

Thin Feature Detection module is used to detect and highlight thin features in the geometry that are difficult to manufacture in AM Process. Thin features are prone to increased thermal deformations due to build up of residual stresses. The thin feature dimensions depend on laser diameter or nozzle diameter. This tool can rotate a part and highlight critical thin sections in the geometry and calculate the areas of thin section region in each layer.

DFAM - Sharp Corner Detection Tool

Sharp Corner Detection Tool is used to detect and highlight sharp corners in the geometry that are difficult to manufacture in powder bed AM process. Sharp corners are tips within each layer where the angle of the tip is less than threshold value of the AM machine. If the angle is too small, these sharp corners in the geometry may be difficult to manufacture during the laser sintering process. This tool can rotate a part and highlights critical sharp corners in the geometry and calculate the number of the sharp corners in each layer.

DFAM - Thin to Thick Transition Detection Tool

Thin to Thick Transition Detection module is used to detect and highlight thin to thick transition areas in the geometry that could cause high thermal distortion in AM Process. Thin to thick transitions are transition areas from a low area layer to a high area layer along the build direction. In the DMLS process, the laser melts the metal powders and fuses them together. The heat energy in the current layer is transferred to the substrate through the layers beneath it. If the transition area between any consecutive layers is too small, it may cause thermal distortion due to impediments to heat transfer during DMLS process.

DFAM - Thin Walls Detection Tool

Thin Walls Detection module is used to detect and highlight thin walls in the geometry that are difficult to manufacture in AM Process. Thin wall is vertical wall structure where the thickness of the wall is too small. Thin walls are very challenge to manufacture duo to the laser spot diameter or FDM nozzle size. Moreover, thin walls could lead to failure or high thermal distortion due to the thermal stress with in the thin wall region.

DFAM - Recoater Arm Collision Detection Tool

Recoater Arm Collision Detection Tool is used to detect and highlight the potential deformation areas or sharp edges of the parts that may collide with the recoater arm and damage the part and/or recoater. Recoater arm spread the powder from one side of the build platform to the other. If there is a long-edge parallel to the recoater, it will be more difficult for the recoater to pass over it in case of deformation. This tool can rotate a part and specify the recoater direction and identify areas of the part that may be prone to recoater damage.

Detection of DFAM features such as small openings, thin regions, sharp corners, thin to thick transition, recoater arm collision and thin walls.

Detection of DFAM Features in Early Design

Accessibility and Setups Analysis Tool

Accessibility and Setups Analysis Tool is used to analyze effect of build orientation on removability of supports as well as the need for setup analysis. The tool can identify support structures that will be accessible/ inaccessible based on user specified directions of tool approach for support removal. In addition, the tool can also optimize sequence of tool approach directions to optimum number of setups for removing maximum support structures.

Support structure accessibility.

Support Structure Accessibility

Producibility Index Calculation Tool

Producibility Index (PI) Calculation and Orientation Optimization

Producibility Index (PI) Calculation and Orientation Optimization

Producibility Index (PI) is the measure of goodness of a design’s manufacturability in a build orientation. This consolidated tool brings together the quantified outputs of the DFAM analysis, support structure parameters and accessibility for the given part geomertry, calculates the PI values of each candidate build orientation uisng weighted optimization scheme and suggests the best build orientations based on highest PI value.

Showing an example of DFAM - Producibility Index Calculation.

DFAM - Producibility Index Calculation


Headshot of Sam Anand

Sam Anand

Professor, Director - Siemens PLM Simulation Technology Center , CEAS - Mechanical Eng

697 Rhodes Hall


Research and Teaching Interests: Intelligent Design, Manufacturing Systems Optimization, Modelling and Optimization of Additive and Subtractive Manufacturing, Industrial Internet of Things (IIoT), Augmented Reality

PhD Students

Headshot of Sourabh Deshpande

Sourabh Deshpande

Headshot of Botao Zhang

Botao Zhang

Headshot of Shailesh Padalkar

Shailesh Padalkar

Headshot of Dorsa Rezayat

Dorsa Rezayat

Headshot of Sirisha Polisetty

Sirisha Polisetty

Headshot of Emily Piatt

Emily Piatt

Masters Students

Headshot of Anuj Gautam

Anuj Gautam

Headshot of Snehita Kilari

Snehita Kilari


  • National Science Foundation
  • MxD (Previously DMDII)
  • Siemens
  • Raytheon Technologies
  • Eaton
  • Intel
  • GE Global Research Center
  • GE Power and Water
  • GE Aviation
  • Procter & Gamble
  • Honeywell
  • International Technegroup Inc.
  • Faurecia
  • TechSolve
  • US Department of Labor
  • City of Cincinnati
  • Dow Chemical Company
  • General Motors
  • Hewlett-Packard
  • University of Illinois at Urbana Champaign
  • Lawrence Livermore National Laboratory
  • Rockwell Collins
  • Optomec Inc.
  • International Trucks
  • Mound Laser and Photonics Center Inc.
  • Ford Motor Company
  • Lexmark
  • Autodesk
  • Ohio Learning Network
  • Cincinnati Sub Zero
  • Kroger
  • Kutz Kasch
  • Mac Tools
  • Health Alliance
  • Sumitomo Electric Industries
  • LensCrafters
  • Convergys