Keywords

1 Introduction

In pipeline construction, especially in repair, fitting pipes play an essential role. The manufacturing steps are mostly carried out manually. The challenge with robot-assisted automation is the small batch size, with mainly one-offs being involved. Since the selection of standard components is defined and recurring components are used, robot-supported automation is nevertheless possible. To be able to integrate this profitably into the entire manufacturing process, the upstream and downstream processes must be considered in addition to the welding process.

For an economic integration of a robot-based solution, the weld seams have to be produced with a consistent quality. To achieve this, the upstream processes have to be adapted. In this case, it is advisable to collect data of the pipe geometries during the manufacturing process and to integrate it into the path planning. In this way, even complex unique structures can be welded with a robot.

In best case the CAD data are integrated directly into the manufacturing process to implement a follow-up of the data and the quality. The digitalization of pipes is already available on the market in various design software packages [1,2,3]. These are based on the new planning of plants and are therefore unsuitable for integration on the construction site. A direct and intuitive integration of the technical data into the entire production process is not practicable. Moreover, additional measurement equipment is required for a detailed representation of the plants. In the case of pipe rehabilitation on ships, this is unsuitable due to the time-consuming transport, limited accessibility, and varying light conditions. Digital Mockups (DIM) are essential in the steps of planning, construction and accessibility analyses. They are used especially in aircraft construction [4], where a scan of the fuselage is carried out and transferred to CAD. Furthermore, there is the possibility to build up the DIM from CAD data. Another advantage of these mock-ups is the comparison of target geometries (CAD) and actual geometries. Due to the high effort, these procedures are mainly carried out for large batch sizes. Due to the lack of documentation, the long journeys and very small premises, scanning the pipes is very work intensive and is therefore not carried out.

A direct integration of the technical data already on site is necessary to improve the quality and the expenditure of time during the entire process chain [5]. Due to these deficits, a methodology was developed which enables the integration of CAD data into the process chain on the construction site.

Currently, a rough hand sketch of the pipe construction is made at the construction site and photos are taken for documentation purposes. After this, a technical sketch is derived based on this manual sketch, which is then transferred to the CAD workstation. The CAD design and a list of parts are derived to order the components. Due to the repeated drawing and changing staff members, errors and mismatches can occur, which can lead to deviations in the required fitting tube. Since the CAD model is not directly integrated into the process chain, these deviations aren’t detected until the end of the process chain. This fact leads to an increased effort in reworking.

2 Approach to Automation

To be able to realize an automation of the manufacturing process by means of a robot, the entire process chain, the manufacturing process and the material spectrum has to be analyzed. Here, the welding of the pipe segments, which has been done manually, has to be optimized by a robot. To achieve the required accuracies for this step, the upstream process steps have to be analyzed and adapted if necessary.

2.1 Analysis of the Manufacturing Process

In the case of one-off production, as is the case in the production of fitting pipes, an exemplary the successive process chain includes the following work steps:

  • cutting the individual pipe segments

  • seam preparation at the pipe ends

  • pre-positioning of the pipes in relation to each other

  • spot welding as preparation for welding

  • welding the joints with a robot.

The main part in the process chain is taken up by the programming of the welding robot. Therefore, it’s important to implement a fast and intuitive integration of the used components in the postprocess and to counteract time-consuming programming. When integrating standard parts like pipe bends, branches, or flanges, predefined libraries should be used to reduce the programming effort, which can be extended and adapted afterwards. Due to the complexity of the components an offline based program is defined through individual CAD data.

To realize welding automation by means of a robot, the quality of the upstream manufacturing steps have to be improved. These production steps include cutting, pre-positioning, seam preparation and tacking. The direct integration and use of CAD data guarantee the required quality.

2.2 Analysis of the Material Spectrum

For the realization of automation, the material spectrum must be considered. This term includes the geometries and the materials. Figure 1 shows pipe constructions, which are used for repairing work on ships. Standard components with individual pipe lengths are used.

Fig. 1
figure 1

Examples for pipe components

For the welding process, the material and the weld preparation are equally important. The most common joints are butt joints. In addition to these, there are pipe branches in various configurations. For this, offline path planning is indispensable, as manual programming of path support points is too time-consuming. The use of optical sensors in online programming can lead to problems with triangulation due to the different angles.

2.3 Analysis of the Production Chain

The manufacturing chain can be divided into three main sections:

  • The upstream manual processes,

  • the partially automated joining processes and

  • the downstream processes.

The upstream processes include the determination and acquisition of the geometry data on site. In the partially automated welding process, the steps described in Chapt. 2.1 are considered. The downstream processes include sawing, surface treatment, straightening and quality control.

The upstream processes are time-consuming and offer a lot of potential for savings. Likewise, this process has a great influence on the final quality.

2.4 Derivation to Automation

In the previous chapters, the current state was explained and the process steps that have potential for optimization were shown. The resulting information is now to be used to derive a digitalization. The digitalization should be designed in such a way that a direct communication between the recording of the geometry data and an automated solution is possible. A functional concept is to be developed from the previously derived requirements.

It has been found that for a robot-based welding concept in a one-off production, offline path planning on CAD data is the most sensible solution due to its high flexibility. However, the generation of CAD data can be error-prone due to multiple processing steps and thus has a major impact on automation. To avoid these multiple steps, the geometric data should be recorded in final form on the construction site. Since the components used for fitting pipes are standardized, the use of a mobile operating device with a database of the standards component is the obvious choice. Figure 2 shows a comparison of the conventional process and the automated process. A specially developed software allows different processes to be combined on a smartpad.

Fig. 2
figure 2

Comparison of the conventional to the automated process chain

The smartpad takes over the creation of sketches, the taking of photos and notes, the recording of geometries and the different components with their specific values. The use of a digital solution eliminates the duplication of drawings. Another advantage results from the fact that all elements are already defined on site and misinformation is minimized by simplified sketches and parts lists.

This makes it possible to integrate the CAD workstation directly into the production chain and reduce the process throughput time. By integrating the CAD data, further process steps can also be partially or fully automated. In addition to automated welding, these include cutting and seam preparation of the pipes.

Due to their structure, programs for documentation purposes have programmatic hierarchies. These can be used to create a defined data structure. Through a targeted query of customer data as well as order data within the program structure, the projects can be structured and systematically stored. This systematic approach simplifies the traceability of data and increases quality. An important step for digitization is the definition of the most important information.

The CAD data generated this way is then used for path planning of the robot. A simulation tool is used to generate this data. In this tool, the seams can be selected, and the required paths are generated. Thus, collision control and control of the welding paths are already possible. The next step is the transfer of the generated welding paths to a robot cell.

3 Implementation of the Developed Process Chain

The developed concept was integrated into the existing process chain. Two different programs were developed for the integration. The first program is an app on an Android tablet, where the technical data of the designed pipe constructions are entered. Ideally, the app guides the operator systematically through the raw design. These information are used in the production chain to optimize the process. For this purpose, a second program is used, which is integrated in a CAD tool. This program creates 3D models and technical documentation from the entered data. In order to test the application possibilities of the developed software, a field test was carried out and evaluated.

3.1 Operating Concept on the Smartpad

To implement the digitisation of the pipe constructions already on the construction site without additional devices, a software for a mobile operating device was developed. This enables an operator to create a pipe construction from several pipe segments. During the input, the pipe construction is displayed according to DIN EN ISO 6412-2 [6] (see Fig. 3). This type of representation allows the operator a standardised visualisation. A database of standard components is stored in the software. With the help of this database standardised components plus their technical parameters can be entered for the fitting pipes. The complexity and size of the fitting pipes are not limited by the software.

Fig. 3
figure 3

User interface for digital geometry input

To create a pipe, the starting point must be entered in the input field for the start coordinates (7) and the end point of the respective pipe segment must be entered in the input field for the end coordinates (8). The pipe diameter and the desired wall thickness must be selected in the control panel (3). Then the weld fittings can be defined in the selection window for weld fittings (flanges, heads, tees, and reducers) (6). The program automatically selects the correct fittings for the selected pipes. Then the segment is added to the current pipe via the “Hinzufügen” (“Add”) button (2). The pipe segment is listed in the table (4) and shown in the drawing (5) according to the standards. To enter connected pipe segments more quickly, the corresponding start option can be selected by choosing the input mode (9). There are several options that make more complex entries more user-friendly, e.g. to connect a pipe segment to a T-piece. Additional information, such as surface treatments or pipe shapes, can be specified using the input fields for additional information (10). Additional design options and layout options are available in the option-buttons (1). New projects and sub-projects are created via this button. There is also an option for notes and photos. The advantage of the user interface designed in this way is the intuitive usability using the real occurring workflow.

After all the necessary data has been entered as described above, the software creates a parts list. This can be displayed and edited. The software always selects standard components for creation. However, if other components are needed in the project, they can be added or changed in the parts list. Figure 4 shows the editing of the parts list using the example of the pipe bends. Finally, the parts list is saved as a CSV file. The parts list is intended as a source of information for the following software in the CAD tool and is therefore not usable without further processing.

Fig. 4
figure 4

Display and editing of the parts list

3.2 Integration into the Process Chain

The aim of digitalization is to increase productivity and quality by capturing the pipe geometry already at the construction site. After the pipe construction has been created with the mobile operating device, the software creates a project-specific folder in which all images, the CSV file and notes are assigned to the construction. The folders are assigned to defined customers and projects and can also contain sub-projects. The automated creation of the folder structure allows for easy traceability and editing. The prepared CSV file can be loaded into a CAD tool by saving a 3D-model and technical documentation such as parts lists and material lists with all the required information. Figure 5 shows the 2D visualization in the app and a 3D view in a CAD tool.

Fig. 5
figure 5

Creation of a 3D view and technical documentation from a technical drawing

The conversion into a 3D model takes place automatically in the CAD tool. The standard components used are defined by libraries. The lengths of the individual pipe segments are formed depending on the total lengths entered and the specific weld fittings. After creating the 3D model, the required pipe lengths and a quantity structure are exported to the operator.

The 3D model can now be integrated into an offline path planning. The MotoSIM tool [7] from Yaskawa was used in the trials. The individual seams can be selected in the tool and the set parameters (torch angle, angle of the positioner, approach, and departure paths, etc.) can be checked. Afterwards, the check for collision and reachability can be carried out. An example of such a setup can be seen in Fig. 6.

Fig. 6
figure 6

Path planning on a digitalized pipe using offline programming

3.3 Evaluation

In the methodology described, a digital concept was developed that shortens an existing process chain and improves the data exchange of individual processes. The digital solution has already been tested in field trials as part of a project. Simple pipe connections can be created intuitively and quickly via the tablet.

During the field tests, a reduction in processing time was observed using the software. Especially the representation according to DIN EN ISO 6412-2 [6] showed great potential in the field tests. The pipes to be replaced could already be matched with technical drawings here.

The field tests were realised up to the CAD workstation. The pipes could be transferred into a 3D model without any incidents. The CAD tool used for the verification was AutoCAD [8]. In AutoCAD, step files could be automatically generated and output, which could be integrated and processed in offline path planning programmes and in CNC processing programs. In the development phase only tubes in thin sheet metal were processed. The generation of the 3D models still required a lot of computing power in some cases.

Challenges arose in the representation of pipe nodes. The intersections between the pipes could not be modelled correctly due to complex interdependencies. Due to missing dependencies between individual components in AutoCAD, the pipe cannot be represented isometrically.

All in all, it could be determined during the field test that the upstream work steps can be automated in an economically efficient way.

Integration into the simulation software MotoSIM was carried out under laboratory conditions. During the tests with the simulation software, the complexity of the seams and the pipes became apparent. The paths created by the software had to be checked and adjusted frequently for the appropriate welding position. The problem here was that a two-axis positioner was used instead of a three-axis positioner.

4 Conclusion

It could be proven that the digitalization of the fitting pipe production is already possible on the construction site. This is done without additional measuring equipment. Automation saves time by eliminating additional individual steps and increases quality by improving the traceability of the data. The data generated by this method can be transferred to various CAD tools.

The next steps are the continuation of the digital process chain, see Fig. 3. Currently, the steps up to the technical documentation have been considered. The transfer of seam preparation for automated welding will be a further step. The model data can already be entered into conventional CNC and path planning programs. However, a comparison between the actual and target model is desirable. The pipes are subject to manufacturing tolerances and have a conicity and wall thickness deviation [9]. This results in new challenges for the automated welding process.

An automated planning of the seam preparation for the thick sheet area has not yet been implemented. The integration of an automated seam planning would be a further optimisation possibility. When calculating the individual intersection contours, the different angles of attack, the tube diameters, the material thicknesses, and the seam preparation must be considered [10].

The integration into an offline program was tested. The next step here is the transfer of the welding tracks to a real robot cell.