What do we Gain with the Digital Twin Technology?
The special digital models, which we call “digital twin”, can be seen as one of the advantages offered by the developments in technology. Under normal conditions, problem detection, studies for productivity enhancement, and decision-making mechanisms were based on real machines, tools, and human data found in the physical environment. While even collecting detailed data on these issues was limited in most of the enterprises, setting the “smart decision-making” process and building data-based methods could mislead enterprises in certain points. Because, unlike an analysis environment where the entire ecosystem can be viewed, the studies were made by using data focused on certain points. The advantage that comes with the digital twin is that the environment and the machine (or even the entire plant) can be moved to a digital environment and data can also be generated in the physical environment and work on different scenarios. Delivering unlimited optimum solutions especially by adding the variable data (weather conditions, exchange rates etc.) to the model and placing these to the top of processes such as production, sales, and planning, creates a high return rate for the enterprises. It is stated that about 500 thousand digital twins are followed in the GE factory in Turkey. In their 2017 predictions, consulting company Gartner, stated that within three to five years, billions of objects will also have a digital twin.
What is Required for Digital Twin Use?
As mentioned in “Product Life Cycle Management” book of Michael Grieves, who is among the first users of the concept of “digital twin” that has been in use since 2001, there are actually three basic factors for building a digital twin model.
The first once for sure, is the device/tool/machine or factory -if we think big- to be twinned in digital environment. Secondly, it is necessary to obtain data that will enable it to move to digital environment. If we want to move the physical environment to digital platform, some points should be put into Mathematics and some data should be translated into computer language of 1’s and 0’s. At this point, sensors lend a hand. The data from the sensors are combined with pre-generated data or with the data already been transferred to digital environment due to integration with several systems, and they get ready for digital modeling.
In order to be able to build the digital twin model, the last step is creating a twin in the digital environment. At this stage, rising methods in computer science such as simulation, prediction, and artificial intelligence come into play. While the digital twin, which is constantly in communication with the physical twin, processes the data and produces results in different scenarios, does not miss any real data that can come from the physical environment, and if necessary, updates its own work based on the new real information.
Although digital twin modeling provides many benefits in important points such as pre-detection of problems, efficiency improvement, and production planning in the most profitable way, if it is only a perfunctory project, it becomes a profitless investment.
Some Errancies of Enterprises Before or During Implementation:
Assuming the implementation is manageable with existing human resources:
For sure, the companies that can acquire knowledgeable staff who can recognize, implement, and manage these new technologies are ahead of the game. If the company does not have employees with this qualification, first, they may prefer to overcome the problem with education; but at some point consultants can be outsourced or new resources can be procured for the team.
Limiting the application to operational contribution:
Although companies, which approach to digital twin concept with this perspective, have a negative effect in practice, as they cannot make use of full potential, they consent to less than they can get. The product/service development processes should be handled through different scenarios on the digital twin, and result of digital twin implementations of different systems should be sent.
Collected data for implementation does not cover the entire loop:
This can be challenging in some projects, which are generally unplanned or limited due to lack of investment. At this point, if the entire loop is not followed in the digital environment, using the values in real world and obtaining correct results becomes more difficult.