Interview with Sergey Lozhkin. Digital twins and simulation modeling

Why have you decided to invest in this direction?

When we have started modeling four years ago, it was not fashionable yet. We rather liked the idea itself. Now digital experiments with digital reality are what awaits us in all areas. Therefore, you need to go there anyway.

For instance, Americans model entire military battles: deploy tanks, planes, infantry, and watch the result of the battle.

In the civilian sector America and Europe have also been modeling for a long time. China is striving for modeling by leaps and bounds. For example, the German company SimPlan simulated an operation of Airbus aircraft using AnyLogic. Mercedes also actively uses AnyLogic. Almost any large company works with simulation modeling.

There are also leading companies that do this in Russia. Now there is a focus on digital transformation: not only in commercial but also in state-owned companies. Today digital transformation is one of the leading topics.

So far our main customers in Russia are commercial entities or commercial entities with state participation – up to 50% of revenue comes from such companies.

What is interesting to model?

It is interesting to simulate processes that are expensive experimentally. For example, you have a huge furnace with melting high-carbon chrome. Any mistake in changing a technological process is a huge expense. Therefore, despite the fact that the efficiency of the process can be increased, you will not go to such an experiment.

In this case, you can create a "digital twin" where to take into account the processes in the furnace and everything else – warehouses, cranes, etc. and simulate the whole thing. For example, you can see what happens if we do not lower the temperature in the furnace.

What is the difference between a "digital twin" and simulation modeling?

Simulation modeling – the process of simulation something. Usually some kind of physical process is imitated. It can be a business process, a process of routing calls, railway and car traffic, logistics processes. By simulating these processes in a digital environment we are building a so-called "digital twin".

That is, simulation – a process, and a "digital twin" – the result that we built in a simulation environment.

What is the difference between simulation modeling and other types of modeling?

The most popular modeling system in the world is MS Excel where we constantly model something. For example, a budget. There is a simulation when we put into the model a function that shows what happens when we conditionally multiply "a" by "b". A function can predict the result right away. The whole problem begins when all the functions in Excel become multi-storey, when the processes become complex, and for each agent the rules are prescribed according to which it can work (go back and forth, left and right), and for calculating it is necessary to calculate the behavior of a hundred objects.

In mathematical functions the result is understandable on the entire horizon. On the contrary, in simulation models the result at the end is incomprehensible. A "digital twin" should live a life in a virtual environment.

What do you mean by "digital twin"?

A lot of things can be adapted to the topic of "digital twin". There are very narrow definitions of a "digital twin", for example, Excel. On the other hand, there are very broad ones when even the introduction of a system like 1C can be called the introduction of a "digital twin". RPA technology is also a "digital twin". This is a very broad concept but we narrow it down to the level of using "digital twins" as twins of a certain physical reality: images of railway transport, cars, etc. This is more understandable on the market.

Hype on "digital twins" in Russia

It is just the beginning in Russia. Now we are receiving more and more real requests, while some time ago really unrealistic requests arrived. For example, they asked us to build a "digital twin" of the entire enterprise which is impossible in principle. This does not solve any functional problem. Now companies are more and more aware of why this is necessary.

A large simulation market is located in the United States. In Russia the volume of use of the AnyLogic product is less than 10% (of global demand). In Europe the construction of infrastructure facilities is impossible without the use of simulation as it significantly increases production efficiency.

Now in Russia simulation modeling is beginning to be used in the design of pedestrian crossings – where to put traffic lights, with what frequency they should work. For example, a "digital twin" of Leninsky Prospekt with all the traffic lights is put into the modeling system. Then they look at how to make optimal traffic with such a stream of cars. Such calculations on paper are simply impossible. Millions of experiments are needed. In real life it would take a lot of time.

Employee resistance

Resistance is most often found where old-school people sit. They hold onto their seats and say that modeling cannot help them increase efficiency. It happens that management is for this idea but we have to work with the masters, dispatchers. There is also resistance on their part.

But now the shift trend is noticeable and it is more and more clearly felt. The "old people" are leaving and the young ones are already thinking differently.

Science and Commerce: Effective Collaboration

The topic of modeling is science-intensive because it uses complex mathematics, different theories. For example, when we were making a project for one of the banking startups, we imitated company's behavior depending on its current cash flow: will it go bankrupt in six months. We used the theory of restrictions. For this, it is necessary to work with the academic community because there are people who are advanced in this topic in terms of developing algorithms and developing the idea of a "digital twin".

Science is the source of ideas for our models or we are doing something together. We are often hired by state institutes of large companies involved in large-scale infrastructure projects. For example, there are specialized institutes within space or metallurgical enterprises that deal with large-scale infrastructure projects, such as the reorganization of production. There are projects from these institutes that subcontract us so that we model their models. Sometimes a business itself wants to check whether their institution has made the right model. Nevertheless, the scientific part is actively used in this direction.