Digital Twin — Construction through Operations
25 May, 2023 By: George BroadbentVIEWPOINT: Digital twins bring data alive in all aspects of a building, from construction to operations and through lifecycle analysis.
What is a Digital Twin? The context matters, but fundamentally the Digital Twin consists of a virtual model, data from an external source that is either real-time or near real-time, and a method for measuring current conditions. We use this construct to adapt and define use cases for numerous scenarios. In this piece, we’ll explore three use cases: Construction, Operations, and Lifecycle Analysis.
The Digital Twin Ladder. Image source: Microdesk.
To understand a Digital Twin, we have to put it into context. A Digital Twin, by definition alone, is either a simple representation of the real world or complex interactions with predictive analytical engines. To visualize this landscape, we’ve included a Digital Twin Maturity Ladder above. In this ladder, Level 01 is the digital artifact or virtual model representing the real world and Level 05 is where predictive analytics and AI models are deployed to help understand the large sets of data being consumed by the Digital Twin.
Digital Twin in Construction
The construction phase can benefit from a Digital Twin in two key areas: installed material and future placement, both of which inform the overall construction progress. The installed material can be further broken down into two subcategories: on-site and installed. Future placement looks at the materials/components yet to be installed at the site. The timeline of events has a graphical, i.e., Digital Twin component and represents a series of use cases with their own set of stakeholders.
Construction digital twin use cases. Image source: Microdesk.
These use cases denote 80% of the objectives for the construction industry. In each of these scenarios, the as-designed model is the reference point for all changes, allowing the team to see how and where the as-constructed conditions deviate. An example of how the planned vs. installed views would look in a Digital Twin are shown below. Additional use cases would include:
- Safety Analysis
- Energy Usage Simulations
- Cut and Fill Estimate Comparison
- Bid vs Installed Analysis
- Pay Requisition Validation
![]() | ![]() |
Planned vs constructed views. Image source: Microdesk.
Through the use of the graphic filters, we can show the various stages of construction in the model considered a Level 1 Digital Twin.
A majority of the data obtained for the Digital Twin in construction comes from unstructured data sources such as equipment specifications, and submittals. The unstructured data can be managed through Commissioning Phase (Cx) platforms to help alleviate the manual data entry process. This is the opposite of what we see in the operations side where the majority of the data is structured.
Digital Twin in Operations
When construction is finished, the goal is (or should be) to have a complete as-built (virtual) model of the constructed facility. This virtual model can then be leveraged as a Level 02-03 Digital Twin. At this point, we start incorporating operational data into the Digital Twin platform, which can be in the form of Real Estate Data, Maintenance Data, Inspection information, Building Management System (BMS), or individual sensors.
How is a digital twin built for operations? There are multiple data sources that come together with the virtual model to form the Digital Twin. These data sources have to be aligned to the physical environment. For example, your Building Management System (BMS)/Building Automation System (BAS) should have equipment and room naming conventions that correspond to the virtual model and your Real Estate and CMMS platforms. This data cleansing phase is arguably the most complex and will take the longest amount of time.
Next is the data collection phase. We have two options when working with a Digital Twin, direct query, or data aggregation. Direct query will be the simplest to deploy, but as your Digital Twin scales, performance will decrease. The long term play here is to have a data aggregation point for near or real-time data. This data aggregation point can be a traditional data warehouse (more expensive, less flexible) or a modern data lake. The data lake will allow for the combination of structured (data) and unstructured (documents) data within one environment that can be referenced by the Digital Twin.
Digital Twin in Facility Operations
Once the Digital Twin environment has been built, the number of use cases is as extensive as the types of data contained. We can summarize the use cases into the following groups:
- Real Estate/Construction
- Director of Facility Operations
- Facility Operations Supervisor(s)
- Facility Operations Planner
Facility operations digital twin use cases. Image source: Microdesk.
Each of these groups have overlapping categories which are explained above.
Lease Holds | The class of digital twin views characterized by occupied/unoccupied space, rental costs (aggregated). |
Construction Progress | One of the digital twin views from the construction phase that allows the stakeholders to view progress of equipment installation. |
Customer Experience | Defines customer as both internal and external groups that are served by the organization. Includes uses such as room temperature, open service requests and overall customer happiness. |
Energy Usage | Defined as overall energy usage and as permitted energy use by equipment. This can be mapped based on the overall facility, space, or equipment. |
Equipment Life Cycle | Digital twin that highlights the remaining life for equipment along with operational costs. |
Operations Cost Analysis | Cost analysis takes into account total labor expenditures and allows for the breakdown of cost by equipment, space or facility. |
Work In Progress | Highlights areas of the facility where there are open work orders or service requests. |
Work Scheduled | Any work orders scheduled for execution over a specific timeline. |
New Work Requests | Requirements from customer groups, typically called Service Requests. |
Shutdown Analysis | Identify upstream and downstream connections for selected equipment, e.g. Electrical Panel / Circuit Breaker. |
Timeline for a Mature Digital Twin
Development stages 01- 05.
There are many industry examples of a digital twin, including cost-benefit analysis and return on investment (ROI). In general, the industry reports a 10% savings on operations and an ROI that ranges from six to 18 months. Since the benefits have been well established, below is a recap of what it takes to develop a fully mature Digital Twin.
Level 01 | Two types of data will be captured in this stage: Geometry and Metadata. The former accurately describes the geometric properties of the building, and the latter focuses on collecting the descriptive information for the elements within the model. |
Level 02 | Data integration connects data from the delivery stage, e.g., the Level 01 virtual model, with data from the operational stage, e.g., Real Estate, CMMS or IOT readings. |
Level 03 | Data is aggregated in a centralized manner, e.g. data lake, for processing. The data is manipulated into a standardized format across all data inputs. |
Level 04 | This is the transition from existing Preventive Maintenance (PM) to Predictive Maintenance (PdM). PdM combines IoT sensor and Machine Learning (ML) technology to enable prediction of the asset condition with high confidence. |
Level 05 | In addition to the data science technologies used in stage 4, more state-of-the-art technologies would be introduced to collaborate for building the AI system. As a prerequisite, all data sources and services collected, prepared, and processed during previous stages now should be connected. On top of that, decision-making optimization, natural language understanding, and other technologies will be architected, modeled, connected, and operated as a virtual assistant. |
Level 01 requires a basic virtual model of the facility that shows all the major components and architectural features of the building. This depends on a coordinated effort on the part of the design, construction, and operations team to ensure that the model represents the facility conditions at the end of construction. It expects the institutional owner will be able to manage the Day 2 changes and the process for updating the virtual model.
Levels 02 – 04 depict a data driven Digital Twin that incorporates data from multiple sources and starts to incorporate ML to create a consolidated and predictive model of current conditions.
Level 05 is the stage of a fully interactive Digital Twin, whereby we can use ML and AI Assistants to help inform and troubleshoot issues within the facility.
As the overall implementation timeline for Level 01 – 05 is not an exact science, therefore ranges of time need to be provided, which are ultimately affected by factors such as Data Completeness and Accuracy. The table shows the progressive timeline of development:
Image source: Microdesk.
This can also be represented as a cumulative timeline shown below.
Cumulative Implementation Timelines. Click image to enlarge. Image source: Microdesk.
Conclusion
A Digital Twin has many use cases and benefits for both construction and operations. Much of the development of its virtual model occurs during the construction phase, with benefits lasting well into operations. Data management and integration are some of the biggest challenges, but once completed the steps to an autonomous Digital Twin (Level 05) can happen at a rapid pace increasing the ROI for its development.
For Mold Designers! Cadalyst has an area of our site focused on technologies and resources specific to the mold design professional. Sponsored by Siemens NX. Visit the Equipped Mold Designer here!
For Architects! Cadalyst has an area of our site focused on technologies and resources specific to the building design professional. Sponsored by HP. Visit the Equipped Architect here!