A robust Assembly Line Balancing System developed for One of India’s most well-known manufacturers and exporters in their product category. Publicly listed.The project included end to end development and deployment of the solution into the factories with proper planning and management.
One of India’s most well-known manufacturers and exporters in their product category. Publicly listed.
The client has many factories with many assembly lines in each. An assembly line has workstations where a worker sits with a machine or tool and works on the manufacturing pipeline. The assembly line needs to be re-constituted at the start of each run for precise matching with the workload, and then the line needs to be monitored to remove bottlenecks rapidly by tuning the mix of workers in the line. There was a need to monitor the throughput of each worker in real time to ensure optimal balancing of resources on each assembly line, so that bottlenecks are eliminated as soon as they form.
Due to the pressures of short batch runs and just-in-time manufacturing, assembly lines need to be re-constituted every few days at most factories. Therefore the challenges of assembly line optimisation are far more severe here than in quasi-static manufacturing environments and real-time feedback of bottlenecks and productivity are required.
Business Benefits with Our Robust Assembly Line Balancing System
The solution helped the client with increased manufacturing productivity. Our client has reported repeatable and measurable increases in productivity exceeding 10%.
- Identifying bottlenecks: If stages 1 to 10 of an assembly line are working very fast, and stage 11 is working more slowly, this becomes visible in a few minutes in the reports. The line supervisor can then add more workers for stage 11 and improve throughput.
- Slow workers: If three workers are performing the same operation simultaneously on the units in the line and their rates differ significantly, then this becomes apparent in the reports in a few minutes. The line supervisor can then replace the slow worker or add more workers for that stage of the line.
- Transparency and visibility: Since the reports are accessed using a browser, managers in their cabins in the factory and the Production In-Charge in the HO can see reports in real time and the throughput of each assembly line in the enterprise then becomes open to scrutiny by all authorised managers in real time. Transparency brings in pressure to perform; it implicitly rewards good performance and triggers scrutiny on poor performance.
- Production planning: Production planning officers who update daily MIS reports about the status of production of each order can monitor the assembly line reports and get real-time status or EOD status of the production status of each order, because assembly line reports carry annotation about order number and other details. With this system, such data used to be forwarded manually by line supervisors at end of production shift, and would be inaccurate sometimes.
Each factory has a central server with the application and database. Each assembly line has a monitoring computer with a browser-based interface for the line supervisors to program the assembly line and job mix details. This monitoring computer connects through RS-232 to a hub unit, which connects over RS-485 to individual devices (referred to as “counter boxes”) attached to each worker’s table. The central server and the monitoring computers communicate over TCP/IP and Ethernet.
The counter box shows two counters and one button. When a worker finishes a job, he presses the button and the system registers the event. One counter decrements, showing count remaining to reach the day’s target. The other counter increments, indicating jobs completed since the beginning of the shift. The devices communicate with the monitoring computer using the Modbus protocol. Therefore, the monitoring computer and counter-box devices of each assembly line form an independent Modbus network.
The button-press events are received by the monitoring computer and sent to the central computer, where status reports are updated every 60 seconds. These are then viewable by supervisors at the monitoring computers and at any other computer connected to the enterprise network, including the HO in Mumbai.
The success of this project depends on getting industrial electronics products of high quality, and partnering with the electronics vendor to handle all interoperability issues in the development phase. We worked with Renu Electronics, a Pune-based manufacturer and exporter of products in the areas of factory automation, building management systems and home automation. Their major strengths are in engineering, customisation of products, and custom solutions. They have been supplying products to the world’s largest industrial automation majors for many years, meeting stringent international quality standards. They customised one of their products to fit the requirements of this project and provided full support to stabilise the finished system.
Software Design The central computer runs industry-standard Java, Tomcat, and an open-source database system. It receives data from monitoring workstations and updates a table in the database. New values replace old values of button-press counts, thereby making the system fault-tolerant of a missed reading. Reports are generated from this table of readings. Data entry screens allow line supervisors to manage the list of workers (each worker has a record with his name and job type, etc), the mapping of workers to tables on the assembly line, details of the day’s jobs, targets for the day, and other details.
The monitoring computers interface on Ethernet and TCP/IP with the central server and with RS-232 and Modbus with the assembly line’s counter-boxes. These computers run a special daemon written in C (an open-source Modbus library) and Perl (using XS wrappers to call C functions from Perl). This daemon is responsible for all communication with the counter boxes. It initialises the counts at the beginning of each shift and reads values from them every few seconds. It communicates in turn with a daemon on the central server to receive commands and send back readings.
The monitoring computers boot into a custom-built pared-down GUI with just a browser window. The line supervisor uses this browser to check his line’s reports and enter data for configuring his line.
Platform and OS The central server and monitoring computers are all entry-level Intel systems. All computers run Linux. The monitoring stations run X Window and Firefox for their purpose-built GUI.
The central server at each factory is backed by a backup server which can take over all services at short notice if the primary fails. Data is replicated from primary to backup server many times a day.
- Challenges: Deploying a new solution on the shop floor involving industrial automation is a larger challenge than developing it. The deployment was initially done at one factory and allowed to run for more than a year. Problems were found with the hardware and software and were debugged. A continuous training and confidence building activity was kept up. During this process, the management at HO about 1,000 kilometres away connected to the interface over the corporate WAN and actively monitored the real-time reports from the system. They communicated on the phone with line supervisors, pointing out bottlenecks. This helped to demonstrate the power of the system.
- The Initial Version: The initial version of the system was developed with electronics devices from a smaller manufacturer, and there were severe quality problems with those devices. It was decided to discard those devices completely and bring in Renu Electronics into the project. The communication protocol shifted from proprietary protocols over RS-232 to Modbus over RS-485. The devices from Renu Electronics continue to be in use and deliver good service.
- Extended Support to Other Factories: After this initial run, the solution was deployed at other factories one by one. In each factory, the same cycle was followed, but times were reduced because the deployment process had been fine-tuned at the first factory. Three of the company’s factories are in the same city, and this made deployment and knowledge sharing easier. Initially, computer operators were hired to do the data entry in the system because line supervisors were not comfortable handling the system themselves. Later, line supervisors took over the operations.
- On-the-Go Tracking: The most important impetus to push deployment was the productivity gains recorded within a few months of deployment at the first factory. In this industry, manufacturing productivity is measured in units per worker-day. A figure of 12 UWD is considered very good by Indian standards, and 17-18 is an excellent figure internationally. In this context, the system helped the factories push up productivity by 1-1.5 UWD. The customer has shared with us that productivity gains of 10% can be observed repeatable. Our estimate is that the customer’s productivity gains may generate sufficient return on investment to pay for the investment in 12 months.