CONFIDENTIAL CLIENT
How one of Latin America's largest truck and bus manufacturers is using linear programming to make its assembly line more efficient
Optimizing assembly line sequencing with linear programming
One of Brazil’s largest heavy commercial vehicle manufacturers wanted to optimize its production planning to raise operational efficiency and cut costs. Facing challenges such as long manual analyses and hits to productivity, the company adopted a solution built by BlueMetrics. Using optimization algorithms based on linear programming, the platform automated critical processes, cutting analysis time from 4 hours to 6 seconds and improving operational predictability.
Overview
Our client has roughly half a century of operations and is one of Brazil’s largest heavy commercial vehicle manufacturers, with a strong presence across Latin America. It produces a full line of buses and trucks, with gross vehicle weights from 3.5 to 125 metric tons. The manufacturer also developed a complete line for passenger transport, focused on the rural, urban, charter, intercity, and school bus markets.
In a sector where operational efficiency is essential to stay competitive, the company saw an opportunity to optimize its production planning process and meet market demand with speed and precision.
“We were called in to help solve a problem that is decisive for any company in the manufacturing segment: the need for more efficiency in processes, which translates into operational gains and profitability. And for that, the solutions we have been building in AI, machine learning, and process optimization proved to be ideal.”
To take on the challenges of managing thousands of components and vehicle model combinations, and to cut the time spent on manual feasibility analysis, BlueMetrics built a custom solution for this manufacturer. Designed to automate and optimize critical processes, the platform let the plant raise productivity, cut costs, and improve operational predictability.
The problem: how to make operational processes more effective and automated?
As happens with any manufacturing assembly line that handles complex processes and automation, production planning at this manufacturing client faced significant challenges:
- Operational
About 4 hours a day went to manual feasibility analysis, adding up to 80 hours a month of repetitive work that was open to error.
- Commercial
Planning failures could lead to line stoppages, delivery delays, and sequencing that was not optimized, all hurting productivity.
- Financial
The lack of precise analysis undercut the plant’s ability to maximize production capacity and created extra costs in inventory management.
In a context of ever-shorter delivery deadlines, a constant push to optimize processes, and steady supply chain challenges, an automated, intelligent solution became essential to keep the company efficient and competitive. This was an ideal use case for the kind of offering BlueMetrics builds: a client with a clear need, a strong impact on company performance, and a call for an efficient, well-fitted solution that could be implemented quickly and deliver concrete results in the short term.
Solution: linear programming in the service of automating and optimizing processes
The solution was designed to automate and optimize production planning at the plant. The technology uses an optimization algorithm based on linear programming to analyze production and inventory data in real time, for greater precision and speed.
According to Diórgenes Eugênio, Head of Gen AI at BlueMetrics, *“The project had several challenges, mainly because it is a very specific niche with detailed concepts. The early meetings were crucial to the project’s success, since that is where the manufacturer’s team could walk us through the whole process and the plant’s pain points within the sequencing process. After the first phase of understanding the problem and the concepts, we worked to understand the data sources and how we could make the process automation happen. From that point, we already knew how we would approach the sequencing optimization technically, but we still needed to define a few strategies to pull all the necessary information from a table with more than 400 rows and 400 columns. After a few attempts and meetings with the client’s team, we optimized the data extraction, finishing with five seconds to collect and transform the data. With the data structures ready, we applied the linear programming algorithm, which always sought to maximize the number of KNRs produced. After a few iterations, we reached a result that guaranteed the best assembly sequence for the production line. The client team’s involvement was fundamental to building this solution.” *
The foundation of a good project is a reliable, resilient data structure. So we built a data and analytics pipeline whose main source is the Prognose table, widely used at the plant. This table lets operators correlate part number balances with the KNRs in production, which makes it possible to assemble the vehicles according to parts availability.
The extraction and transformation of this data happened in the first stage of the pipeline, converting it into structures optimized for running the optimization algorithm. These new structures provide key insights, such as KNRs pending analysis, the updated balance of part numbers, and the list of components needed to assemble each KNR.
With the information organized, we implemented an optimization algorithm based on linear programming. The algorithm was designed to maximize a specific goal: produce as many KNRs as possible, making efficient use of the available part number balances and the viable assembly combinations.
Finally, the solution was built around three important pillars: automating data collection to make it faster and more reliable; optimizing decision-making so the best production options are presented quickly and accurately; and a friendly, secure, intuitive interface to make adoption easy across the different teams involved.
Main features:
- Automated data collection:
- Optimization algorithm:
- Intuitive interface: Want to see GenAI and machine learning making a difference at your company?
Results:
Rolling out the solution brought strong operational gains for the client, translating into concrete, immediate financial results. On top of that, the proprietary blue4AI method plans for a view of continuous optimization after delivery, so it can spot new opportunities for improvement and positive impact on the business.
- Operational efficiency:
- Production optimization:
- Financial impact:
- Qualitative benefits:
Technologies used
The solution was designed using AWS technologies, including:
AWS services
EC2
Application Load Balancer
Languages, libraries, and frameworks
Python
Streamlit
Pulp
OpenPyxl
Pandas
Plotly
Conclusion
With the platform we built, the manufacturer reinforced its commitment to innovation and efficiency, cementing its place as a reference in the use of advanced technologies for the automotive sector. For BlueMetrics, this case was a great opportunity to show how companies work, applying a proprietary methodology that gives projects real speed and fit, along with an implementation that is quick and uncomplicated and that keeps the focus on business vision and on generating concrete, short-term results for the client.
Want to build a case like this one for your company?Let’s have a chat.