Case study: modeling the work of a pipe rolling factory

modeling the work of a pipe rolling shop

AnyLogic specialist Gennady Parshin simulates production operations at the United Metallurgical Company (OMK) - a world leader in the production of large diameter pipes. In this article, Gennady shares his experience in modeling complex technological processes using AnyLogic and talks about how OMK engineers simulate the operation of overhead cranes. We publish this article on his behalf.

Gennady Parshin, OMK
Gennady Parshin,

In one of the workshops of the enterprise, we installed a new electric pipe mill which, in turn, influenced the production material flow.

Before the construction works in the workshop, work was carried out in two bays. This is how it looked:

  • In the first bay, crane 1 moved sheets of rolled metal from warehouse 1 to trolley 1.
  • In the second bay, crane 4 transported rolls from trolley 1 to the inspection platform in front of the cutting unit.
  • Crane 3 moved rolls from warehouse 2 to the inspection site.
  • From the inspection platform by crane 4, the coils were moved to the inlet of the slitting unit for cutting into strips — blanks for pipes.
  • The strips were moved by crane 2 to the strip warehouse in front of the electric pipe mill, and then by the crane to the mill.
Material flow before reconstruction
Material flow before reconstruction. The flow of rolls and strips is shown with red arrows and boxes of waste are shown in blue.

Waste after cutting the steel coils goes into boxes near the cutting unit. The filled boxes are weighed and unloaded into wagons. Boxes with packaging tape waste are unloaded into the wagon in bay 2 without weighing.

During the reconstruction, a third bay was built in the workshop, where new cranes were launched, and a new mill was installed across all the bays. Because of this, the material flow in the workshop changed. Now, strips from the cutting unit go on trolley 1, then to trolley 2, and then to the warehouse, from where they move to the mill. The cut waste stream has also changed and is shown in blue in the diagram. In addition, planning the work of cranes 1 and 2 became more complicated as they should not work while trains and cars pass under them.

Material flow after reconstruction
Material flow after reconstruction.

The complexity of the new scheme prevented planning work “on paper” or in Excel, so the plant’s management decided to visualize the new configuration of the material flow and test its impact on the productivity of the shop. To do this, I created a simulation model of the workshop and production operations. It reflected:

  • movement of rolls, strips, and waste boxes;
  • movement of carts, cars, and trains in the workshop;
  • incoming control of rolls in front of the cutting unit;
  • operations for cutting a roll into strips.

This is what the shop floor model looks like:

Simulation of the production process of cutting rolled steel coils into strips for an electric pipe mill.

How the model was created

In the model, I tried to accurately reproduce the space of the workshop and the transport equipment. For this, we used ready-made elements from the specialized AnyLogic libraries. They helped speed up and simplify the modeling process, and their 3D visualization made it possible to visually show operations in the workshop and verify layout solutions.

Workshop schema
Workshop schema.
Shop simulation model
Shop simulation model.

Workshop equipment included:

  • transport carts – to transport goods between workshops;
  • trains – to take finished pipes from the shop;
  • trucks – to transport boxes of cutting waste along the tracks.

However, the main transport equipment in the workshop is the overhead multi-girder cranes. They not only move rolls and boxes with cutting waste, but also participate in the placement of machinery on the cutting unit and load-handling devices.

The operation of such cranes in factories can be difficult to simulate due to the complex logic of the movement of the beams. In AnyLogic, this process is simplified by the special ready-made Overhead Crane. It allows you to easily simulate the loading of objects, including from a conveyor network or railway tracks. Working in automatic mode, it distributes tasks between beams, prioritizes work, and resolves conflicts. It also has a manual mode, in which the user can define complex logic for the operation of beams. The number of beams can be configured in one click in the object properties.

In AnyLogic, crane operation is modeled using three main blocks – SeizeCrane (grabs the crane), MoveByCrane (moves the object on the crane) and ReleaseCrane (releases the crane). You can read more about the technical features of overhead crane operation in AnyLogic help or learn from the demo model.

Process diagrams in the model
Process diagrams in the model

In the model, each process diagram is responsible for one of the three crane routes. When an agent enters the SeizeCrane block, it enters the crane’s queue, in this example, craneBridge.

Process diagram simulating crane operation
Process diagram simulating crane operation.

Model hierarchy

Each crane in the plant operates several girders, which in turn are responsible for gripping, moving, and loading cargo. So, simultaneously, and according to the schedule, the cranes perform dozens of unique operations that need to be reflected in the model. A standard approach to creating a model — a common process diagram for all operations — would complicate the model's maintenance: a “flat” diagram would be too long and confusing, and it would be more susceptible to error when changing its component parts.

To avoid this, I used a hierarchical approach that helped structure the model and make it usable. The model is divided into three levels, each of which is responsible for the operation of cranes at different levels of abstraction. At the same time, I divided the entire production process into sub-processes to make the model simpler and easier to reconfigure.

Layered diagram of production process modeling
Layered diagram of production process modeling.

At the lowest, zero, level there are production process objects, including multi-girder cranes with characteristics and parameters for the movement of actuators. Thanks to the capabilities of AnyLogic, the properties of these objects can be easily configured, and the detailing allows you to simulate the system close to reality.

Crane agent and its properties
Crane agent and its properties
Crane agent and its properties.

On level one, I reproduced the processes of production operations (the diagrams of technological operations, DTO) – the movement and production of parts and pipes, and the unloading and loading of trains and cars. They combine objects from level zero as well as repetitive manufacturing operations.

It is convenient to model such repetitive operations as custom blocks. These elements help simplify the look of the process diagram and they can be used multiple times in a model. Check out how I simulated moving the boxes onto a cart: the crane repeats this operation throughout the manufacturing process, and these custom blocks help speed up model creation. In the model, these blocks are called the Elements of Technological Operations (ETO), they are at the first level and are part of the DTO diagrams.

An example of such nesting is in the DTO diagram below. In the Source block, the coordinates of the node where you want to take the box are set, and further movements are defined in the ETO objects.

An example of an ETO agent type: Crane 5 moving a waste box from a position in bay 3 to trolley 2.
An example of an ETO agent type: Crane 5 moving a waste box from a position in bay 3 to trolley 2.
DTO diagram consisting of ETO objects (crane routes): removal of waste boxes from bay 3.
DTO diagram consisting of ETO objects (crane routes): removal of waste boxes from bay 3.

On the second level, production operations are combined into a production process - a top-level representation of the work of the shop. In my model, processes are described as a sequential flow of events in the form of a Process Execution Model (PEM) diagram. It controls the processes at the lower levels, and in it, the agent sequentially walks through each block and launches DTO diagrams, as well as auxiliary blocks. There can be several PEM diagrams, and they can interact with each other.

PEM Diagram: Top-Level View of Shop Floor Operation
PEM Diagram: Top-Level View of Shop Floor Operation.

As you can see, a hierarchical approach to model creation helps make the model flexible. Each level is like a separate, nested model with its own process diagram. Such a nested model is easier for new users to understand because it is not overloaded with unnecessary details, and when its logic changes, the likelihood of the incorrect operation of linked models is reduced. In addition, the hierarchical approach makes it easier to integrate new elements into the existing model.

Here are examples of models with hierarchical logic from the AnyLogic developers. They can be downloaded from the links or found in the program’s example section:

Simulation results

Thanks to the AnyLogic Material Handling Library and the flexibility and detailing of its elements, it was possible to accurately simulate crane operations in the workshop. In doing so, I took into account the changes in material flows, the work of transporters, and the new topology of the premises.

The simulation results helped to determine the productivity of the workshop, plan the movement and cutting of steel coils, and set out the operation of rail and road systems.

One of the main features of the model is its multi-level architecture. Using this approach, it is possible to create other models of multi-level systems, for example, prototypes of an automated process control system. On the lower levels, it was possible to simulate unit prototypes with sensors and actuators, and on the upper levels, the information flows of processes, for example, the transfer of data about the performance of technological operations. This is another very interesting area in which to apply AnyLogic.

Related posts