In the context of digitalization and automation of inspection and processing processes, a special role is played by classification systems elements – a kind of “digital dictionaries” that ensure uniformity in the description and parameterization of objects. Classifiers form the “common language” that allows data to be grouped by meaning and data to be integrated between different systems, management levels and phases of the project lifecycle.
The most tangible impact of classifiers is in the economics of the building life cycle, where the most important aspect is the optimization of long-term operating costs. Studies show that operating costs account for up to 80% of the total cost of building ownership, which is three times higher than the initial construction costs (Fig. 4.2-6) (W. B. D. Guide, “Design for Maintainability: The Importance of Operations and Maintenance Considerations During the Design Phase of Construction Projects,” in Design for Maintainability: The Importance of Operations and Maintenance Considerations During the Design Phase of Construction Projects). This means that the decision on future costs is largely formed at the design stage
This is why requirements from operations engineers (CAFM, AMS, PMS, RPM) should become the starting point for generating data requirements during the design phase (Fig. 1.2-4). These systems should not be viewed as the final stage of the project, but as an integral part of the entire digital ecosystem of the project, from concept to disassembly
A modern classifier is not just a system of codes for grouping. It is a mechanism for mutual understanding between architects, engineers, estimators, logisticians, maintenance and IT systems. Just as a car’s autopilot must unambiguously recognize road objects with high precision, digital construction systems and their users must interpret the same project element unambiguously for different systems via the element class.
The level of classifier development directly correlates with the depth of a company’s digitalization and its digital maturity. Organizations with a low level of digital maturity are faced with fragmented data, incompatible information systems and, as a result, incompatible and inefficient classifiers. In such companies, the same element can often have different grouping identifiers in different systems, which critically complicates final integration and makes process automation impossible.
For example, the same window in a project can be labeled differently in CAD model, estimating and maintenance system (Fig. 4.2-7) because of the multidimensional perception of elements by different participants in the process. For the estimator in the windows category element, volume and cost are important, for the maintenance service – availability and maintainability, for the architect – aesthetic and functional characteristics. As a result, the same element may require different parameters.
Due to the difficulty of unambiguously defining the classification of building elements, specialists from different fields often assign incompatible classes to the same element. This leads to a loss of a unified view of the object, which requires subsequent manual intervention to harmonize the different classification systems and to establish correspondence between the types and classes defined by different specialists.
As a result of this inconsistency, the operational documentation received by the procurement department (ERP) when a construction item is purchased from a manufacturer often cannot be correctly linked to the classification of that item at the construction site (PMIS, SCM). As a consequence, critical information is not likely to be integrated into infrastructure and asset management systems (CAFM, AMS), which creates serious problems during commissioning, as well as during subsequent maintenance (AMS, RPM) or replacement of the element.
In companies with high digital maturity, classifiers play the role of a nervous system that integrates all information flows. The same item receives a unique identifier that allows it to be transferred between CAD, ERP, AMS and CAFM -systems and their classifiers without distortion or loss.
To build effective classifiers, you need to understand how the data is used. The same engineer may name and classify an element differently in different projects. Only by collecting usage statistics over the years can a stable classification system be developed. Machine learning helps with this: algorithms analyze thousands of projects (Fig. 9.1-10), identifying likely classes and parameters through machine learning (Fig. 10.1-6). Automatic classification is especially valuable in environments where manual classification is not possible due to the volume of data. Automatic classification systems will be able to distinguish basic categories based on minimally populated item parameters (more details in the ninth and tenth parts of the book).
Developed classifier systems become catalysts for further digitalization, creating the basis for:
- Automated estimation of project cost and schedule.
- Predictive analysis of potential risks and conflicts
- Optimization of procurement processes and logistics chains
- Creating digital doubles of buildings and structures
- Integrations with smart city and Internet of Things systems
The time for transformation is limited – with the development of machine learning and computer vision technologies, the problem of automatic classification, which has been unsolvable for decades, will be solved in the coming years, and construction and design companies that fail to adapt in time risk repeating the fate of taxicabs displaced by digital platforms.
More about the automation of costing and scheduling as well as big data and machine learning will be covered in the fifth and ninth parts of the book. The risk of a repeat of the fate of taxi fleets and the Uberization of the construction industry are discussed in detail in the tenth part of the book.
Understanding the key role of classifiers in the digital transformation of the construction industry, it is necessary to turn to the history of their evolution. It is the historical context that allows us to realize how approaches to classification have evolved and what trends determine their current state.
