Robotic assembly

Citation

Edmondson, N.F. (2005), "Robotic assembly", Industrial Robot, Vol. 32 No. 3. https://doi.org/10.1108/ir.2005.04932caa.002

Robotic assembly

Robotic assembly

Keywords: Robotics, Assembly lines

Seven years ago I walked into “Q's” laboratory from the latest James Bond film, or at least, that was how it appeared! I was surrounded by teams of people working on strange contraptions with pneumatic cylinders and levers, lasers for cutting and welding sheet metal and robots of all shapes and sizes attached to the floor or hanging from the ceiling. The robots were being programmed and tested at frightening speeds and were hypnotizing to watch. By the side of one of the machines was a sign that read:

• •

  BE THINK INNOVATE

It was my first day at the Grundfos Technology Centre in Denmark, where they develop and produce robotic systems for their global manufacturing facilities I was being shown around by Wolfgang, who could be described as the German equivalent of “Q”.

This was my dream job. I spent the next 4 years developing and implementing flexible part feeders and robotic assembly systems, working with and developing ideas fr om the leading figures in the field. Alan Redford who was a key player in developing Geoffrey Boothroyd's DFA, system and Tony Williams from the INFACT project, which developed a modular generic assembly system, were my guides and mentors.

Denmark has one of the highest labour costs in Europe, which lead Grundfos to conclude that if it was to maintain production in Denmark and other Western European countries, it was going to have to reduce its labour costs. This realisation led directly to the birth of the Grundfos's Automation strategy.

Grundfos's automation strategy has been developed for application when any or all of the following opportunities exist.

1. 1.

   Quality – where the automation of a process will provide clear quality benefits.

2. 2.

   Health and Safety – where automation will reduce the risk to human operators.

3. 3.

   Potential cost reduction.

At Grundfos, robots have been integrated into dedicated high-speed machines to provide economic assembly with high flexibility at mass production volumes. In much the same way as companies such as Sony have automated assembly of their video cameras and Walkmans.

For example, the Grundfos circulation pump, has over 800 different variants produced on the same highly automated assembly line, to suit detailed variations in customer requirements.

To achieve this, Grundfos has developed an assembly system capable of mass customisation, which is built-up from a number of automated stations, linked together using a free transfer system. At each automated station, vision system based flexible feeders are used to feed those parts that are used to create different variants. Part insertion and handling is conducted using a combination of robots and programmable assembly stations. Standard gripping and fixture locations have been identified so that gripper and fixture variants are not required. In this way, the cost of design changes and the introduction of new product variants have been minimized. The modular construction of the line enables new assembly stations to be introduced to increase capacity or to replace the few remaining manual assembly stations.

In the late 1990s Grundfos began to re-evaluate the validity of flexible assembly systems. We identified that due to the increase in labour costs, the reduction in the cost of robotic technology and a significant improvement in the performance of control technology, flexible assembly could offer significant savings over semi-automatic assembly. Based on this conclusion we started a research project, which led to the development of a generic modular assembly system for medium to low production volume products.

The early pioneers of flexible assembly often stated that the cost of flexible part feeding was the major limiting factor for the application of automatic flexible assembly in low to medium volume production. We found that whilst the cost of feeding is still significant, the major cost of implementing this technology lies in the programming and running-in of the equipment, i.e. checking/testing that all of the different sets of assembly parameters can be implemented and that the robot co-ordinates are correct.

A dedicated assembly system assembling only one type of product is significantly cheaper to run-in than a flexible assembly machine. This is because when many product variants and different products are being assembled, each new product and variant needs to be run-in. Furthermore, the running-in process often takes a long time because the full range of component tolerance variation needs to be tested in the machine. When you are producing in small volumes it may take months to see the total process variation of all component batches. This is particularly critical when assembling plastic parts. The result of this variation in part dimension combined with the positioning variation of the robot, gripper and fixtures, is that many fine adjustments in the robot program are required.

But, what about vision systems and compliance devices? We found the use of vision and compliance systems to be an interesting solution to compensate for the tolerance problems. However, they add cycle time, programming complexity and significant cost to our flexible assembly systems. We also investigated the use of strain gauges based compliance systems, which allow the assembly robots to “feel”. However, due to the programming complexity, cycle time and cost, we decided to use a simple mechanical compliance device to compensate for component variation in X-, Y- and Z-axes. After conducting a 6-month production trial, we concluded that we could eliminate the need for compliance in the X- and Y-axes, by simply increasing the part chamfers. This conclusion leads to a different way of thinking.

If we designed our products for assembly (DFA) using standard modules that could be plugged together to create different products and variants (which means standard modular feeders, and programs) and really understood the reasons for our process variation (DFM...Six-Sigma), we would not need to use complicated technological solutions to solve our tolerance problems in the assembly process. Furthermore, the use of standard modular parts would reduce the number of different variants that needed to be run- in. This was the fundamental issue for me.

I realised that by applying DFA, we could drastically reduce the cost and complexity of our product assembly and by using the six-sigma methodology and toolbox, and we could reduce the component variation by identifying and controlling the critical process parameters.

Today, we have secured the foundation of our automation program by ensuring that every new product designed at Grundfos undergoes a team based DFA analysis. We use six- sigma to improve and control all of our critical production processes and we are developing modular products, which can be assembled by robotic-based flexible assembly technology.

In the long term, the catalyst for the wider implementation of assembly robots will be the development of intelligent software and smarter vision and sensing technology, which will simplify and speed-up the programming of these systems. So “Q” still has a lot of work ahead to ensure that James Bond will once again be in a class of his own!

Nigel F. EdmondsonBased at the Grundfos A/s, Bjerringbro, Denmark