First, ML requires a meta data model, i.e.i.e. a data structure for core asset data that is suitable for AI, e.g. ontologies like bricks, haystack or real estate core. Second, detailed data/information about the assets facilitating the O&M process. Third, data measuring the performance of the asset in question, real-time and historical

Formulating the problem mathematically for advanced multivariable predictive analytics is challenging

ML requires a lot of structured data to train and validate the models

BIM models are generally not delivered in as-built condition. Cost focus leads to suboptimizations such as modeling assets in one space and referring to them in others. Assets in the BIM model are suggestions, the actual asset that is installed is not specified. Sometimes this information is delivered separately without any connection to the model or in formats that are challenging to integrate into the AIM.

Creating a structured AIM is resource intensive and the benefits are not instantly accessible. Due to the general lack of structured data the AECO-FM industry has an especially poor starting point

Ideally the 3D drawing space needs to be auto translated into building knowledge graphs as they are easily become too large to be effectively maintained by humans

Interoperability is vital to enable editing, ensuring the as-is/as-built state of the AIM.

The process of updating the AIM needs to be integrated into the daily work processes

Divergence in digital maturity between actors

Generally low digital maturity in the industry

Data and information segregation

The transaction speed of blockchain makes it inappropriate for high data rates, off-chain storage with references on chain could circumvent this issue

Some BCT protocols are energy intensive

It should not cost too much in terms of computing power and performance to decrypt anything that is not of particularly high dignity

Blockchains are difficult to program, cost of implementation could be higher than the potential value

Access control credentials, which credentials has access to what spaces during what time

Who has access to what information in the DT.

Keeping track of service intervals, protocols, inspection logs, pictures, state of assets, active/retired assets

Transactions in general requires an immutable state, e.g. data/information handovers could utilize BTC to establish consensus regarding what was delivered by whom and when

If there is a need for security of a certain type of information blockchain could be useful

The level of encryption should stand in proportion to the level of sensitivity/importance of the information/data. Sensitivity/importance of data can vary between different types of facilities

Recording data and decisions made by AI provides transparency and traceability promoting higher trust and ultimately better results