Taming Data Complexity: Leveraging Tech to Manage Large Datasets

Digital twins contain large datasets, such as point cloud files from LiDAR surveys, as well as 3D design, construction, operation, and maintenance data. Image source: Decastro/stock.adobe.com.

As ever-advancing technology expands horizons for architecture, engineering, and construction (AEC) professionals, the burgeoning amount of data also presents ongoing challenges in how to manage all of this data. With information emanating from multiple sources in increasing volumes, AEC professionals need new ways to process and manage large datasets while maintaining efficiency throughout project lifecycles. Teams equipped with the right workstations and related technology are best suited to handle these complex projects.

Dealing with a Deluge of Data

The information deluge often begins in early project stages, as project surveys conducted with light detection and ranging (LiDAR) and other techniques generate massive point-cloud files. Existing mapping sources, property information, and a variety of spreadsheets, text-based documents, and other digital records add to the volume.

As design and construction work progresses, the data volume expands. Preliminary design work may generate multiple 3D models, renderings, and visualizations. Final design work produces project details to incorporate into models and construction documents.

And for the growing number of projects that feature digital twins — virtual representations of physical products, processes, and facilities — additional data is generated. Digital twins include detailed geometric design data, plus a wealth of information that helps owners manage operations and maintenance activities. Equipment specifications, photographic imagery, sensor and IoT data, and other documents are often incorporated into digital twins to help owners monitor performance of systems in real time.

Data is also generated by AI-driven, digital-twin simulations that predict how systems will perform and when equipment might need replacement. As equipment and other infrastructure are replaced, the digital twin can be updated to reflect current conditions. AI and other automated processes can also be used to manage pricing, scheduling, and other documentation related to operations and maintenance. All of this information provides value throughout project lifecycles, but the volume of data requires ongoing management to maintain access and efficiency.

Digital twins are virtual representations of physical products, processes, and facilities. Image source: NVIDIA.

Solving Data Management Challenges

To manage the large datasets associated with digital twins, AEC teams need computer systems with plenty of horsepower. This generally means high-end central processing units (CPUs) with multiple cores, high-performance graphics processing units (GPUs), and plenty of memory and storage space.

Along with powerful computing systems, AEC professionals also need robust data management systems with high levels of security to maintain data integrity and confidentiality. Workstations such as the Lenovo ThinkStation P-series workstations are well equipped to handle on-premises storage for digital twins and AI-driven projects.

This provides teams with more control and potential security advantages over cloud solutions, which may use third-party servers for data storage.

AEC teams should coordinate additional security measures with IT operations to prevent unauthorized access, data breaches, and manipulation of data flowing between the physical and digital environments. Policies should be established and enforced on data use, data ownership, and password complexity.

Lenovo’s ThinkStation P family of workstations supercharged with NVIDIA graphics cards are well-equipped to process and manage large datasets for digital twins and other AEC projects. Image source: Lenovo.

Collaborating Across Distances

An effective digital twin establishes a single source of truth — one federated model that consolidates all relevant data from various sources and of various formats. Data might include geometric models, mathematical models, simulations, equipment data, operation and maintenance records, imagery, correspondence, and more. Basically, anything associated with the project could become part of the digital twin.

As the single source of truth, digital twins must also be physically accurate with true-to-reality physics, materials, lighting, rendering, and behavior, synchronized to the real world. This enables teams to identify a single moment in time and accurately simulate and predict “what-if” scenarios. Most digital twins are built and maintained by multiple collaborators, often dispersed in different locations. Consequently, digital twins must be accessible by all authorized team members at any time, with data updated in real time. Team members must be able to connect to all project data and extend to IoT, data systems, and industrial automation tools, with confidence that they are always accessing current data.

Lenovo’s ThinkEdge servers and  NVIDIA Omniverse solutions provide reliable distributed processing capabilities, enabling seamless collaboration across disciplines while ensuring data integrity.

USD is file-system agnostic, with data not limited to specific file systems or applications. Image source: NVIDIA.

Bringing it All Together

With the large datasets of digital twins, along with collaboration, synchronization, and security needs, AEC teams need dependable, integrated solutions that combine high-end computing power and advanced technology to manage the demands of digital twin workflows. Team members will likely use a wide variety of applications and generate disparate data, and everything needs to work together.

Hardware needs may vary depending on work types and individual roles, and Lenovo’s ThinkStation P-series workstations are designed to provide a wide variety of capabilities, ranging from manager-level functions to complex data processing and modeling.

For example, the ThinkStation PX, P7, and P5 models feature dual 4th Generation Intel Xeon Scalable processors with up to 120 cores and support up to four NVIDIA RTX GPUs, each equipped with up to 48GB of memory. This combination enables real-time rendering, AI-driven simulations, and seamless manipulation of massive datasets, critical for digital twin workflows. These workstations are also equipped with high-endurance storage drives for rapid data access and up to 2TB of memory, allowing AEC professionals to work with large datasets smoothly. Unlike standard PCs, these workstations are engineered for 24/7 operation, providing reliability and uninterrupted productivity.

For AEC professionals requiring mobility, Lenovo’s ThinkPad P-series mobile workstations offer powerful configurations tailored for working with digital twins remotely.

Facility owners such as Amazon have used Omniverse-based digital twins to visualize warehouse facilities in full-fidelity realism, aiding operations, and maintenance planning. Image source courtesy of Amazon Robotics.

These portable solutions integrate high-performance NVIDIA RTXTM GPUs and Intel Core or Xeon processors, enabling engineers and designers to handle complex 3D models and simulations even while on the go. To complement both desktop and mobile workstations, Lenovo’s ThinkShield provides an extensive portfolio of secure solutions to protect remotely accessible data around the clock.

At the heart of a successful digital twin is an open ecosystem geared for complex 3D operations and interoperability between datasets. Universal Scene Description (USD) provides an ideal ecosystem for digital twins. Originally released by Pixar in 2016 for entertainment applications, OpenUSD has steadily grown across industries working with heavy 3D content, such as architecture, engineering, manufacturing, and scientific experimentation.

OpenUSD enables numerous operations, such as modeling, rendering, animation, lighting, physics, and particle effects, to be represented in the same 3D environment for specific requirements and workflows. Scene descriptions can be created, serialized, and shared across multiple data pipelines. Traditionally, these representations could not be natively understood or modified outside of the original host application. OpenUSD enables multiple representations to be created and integrated, with open APIs available for building custom applications.

To incorporate OpenUSD in a digital twin environment, NVIDIA Omniverse enables developers to build interoperable applications. Built on OpenUSD, Omniverse provides a platform of application programming interfaces (APIs), software development kits (SDKs), and services that enable developers to orchestrate multiple technologies into existing workflows. By creating Omniverse applications to view and interact with digital twins in Omniverse, teams can build virtual replicas of unique objects, processes, and environments, all in sync with real-world data and powered by AI.

Applications built with NVIDIA Omniverse have proven invaluable for facility owners around the world. Amazon Robotics, for example, is building unique tools for their digital twins of their warehouses in NVIDIA Omniverse to better optimize warehouse design and flow, to train more intelligent robot assistants, and gain overall productivity. Pepsico and other companies have undertaken similar efforts.

By leveraging Lenovo’s computing solutions with NVIDIA RTX GPUs and Omniverse, AEC professionals can enhance visualization, improve real-time simulations, and ensure seamless data management across the entire lifecycle of a project. This combination of cutting-edge hardware and AI-driven solutions provides the horsepower required to tame data complexity and fully leverage the potential of digital twins. 

Download this article as a whitepaper here.

*** 

This article was sponsored by Lenovo and NVIDIA. 

 

 

 

Searching for more information about Architecture, Infrastructure, and Construction? 

Click here!