Powering a Next-Generation Digital Redrying Facility with TDengine

Honghe Redrying Factory, part of the China Tobacco Hongyun Honghe Group, specializes in tobacco leaf grading, redrying, and warehousing. In September 2024, it successfully completed a major relocation and technical upgrade, launching a new digitalized facility that features China’s first redrying production line equipped with extended intelligent redrying machines, increasing annual processing capacity to over 33,000 tons of tobacco leaves.

By deploying an integrated manufacturing operations management (IMOM) system and adopting TDengine for real-time, cost-efficient handling of massive-scale time-series data, the company has taken a key step toward digital transformation and intelligent manufacturing.

Background and Challenges

Redrying is a critical link in the tobacco industry value chain, bridging the gap between agriculture (tobacco cultivation) and industrial manufacturing (cigarette production). It is considered the core stage of primary tobacco processing. Highly specialized in nature, this process can be understood as the stage where tobacco leaves are standardized and their quality is refined.

Although raw tobacco leaves from different growing regions undergo initial curing, they still exhibit uneven moisture content, impurities, and inconsistent quality, making them unsuitable for direct use in cigarette production. Through the redrying process, these leaves are further cleaned, conditioned, and stabilized, ultimately producing lamina that meets the requirements of the cut tobacco processing stage. This ensures a consistent, uniform, and stable raw material foundation for downstream manufacturing.

Redrying is a typical process manufacturing operation. While its data characteristics are similar to those of cut tobacco processing, it also has its own unique features:

• High-Density Data Sources with High Sampling Frequency: A redrying production line consists of various types of equipment, including vacuum conditioners, threshing and air separation systems, redrying machines, and pre-press balers. Each type of equipment is equipped with a large number of sensors measuring key parameters such as temperature, humidity, pressure, flow rate, air velocity, and motor current. To achieve precise process control, data is typically sampled at second-level or even higher frequencies, resulting in hundreds of thousands of time-series data points generated per second across the entire production line.
• Strong Time-Series Characteristics with Close Process Coupling: All data generated during the redrying process is strictly time-stamped and exhibits highly continuous time-series characteristics. Various process parameters—such as temperature and humidity variations across different zones of the redrying machine—directly affect key quality indicators of the processed tobacco, including moisture content, color, and aroma. Therefore, it is essential to accurately trace the complete process curve experienced by each batch of tobacco leaves to ensure product consistency and enable process optimization.
• Massive Data Ingestion Pressure: The system must continuously ingest large volumes of data during production, making it a typical write-intensive application. Traditional relational databases struggle under such sustained high-throughput write workloads, with performance degrading rapidly and ultimately becoming a system bottleneck.

Why TDengine

TDengine was selected as the core data platform for the digital transformation and intelligent manufacturing upgrade of the redrying plant. The key reason lies in the strong alignment between the data characteristics of the redrying process and TDengine’s performance advantages, enabling end-to-end support from data acquisition to process optimization.

• Zero-Code Data Ingestion: Direct data ingestion from OPC UA servers is built in to the system. Through the TDengine TSDB Explorer web interface, data ingestion tasks can be configured graphically, with support for dynamic updates of data points.
• High-Performance Ingestion: TDengine creates an independent table for each data collection point and adopts a columnar storage model with append-only writes. This architecture, purpose-built for time-series data, delivers write performance an order of magnitude higher than general-purpose databases, easily handling the full data stream generated by all sensors in the redrying workshop.
• Ultra-High Compression Ratio: Time-series data is inherently highly redundant. TDengine applies optimized compression algorithms tailored for such data. In redrying scenarios, compression ratios typically reach 10% or better, meaning storage space can be reduced by over 90%, significantly lowering the cost of long-term data retention and archiving.

TDengine TSDB Implementation in Practice

To meet the data acquisition requirements of the IMOM system, Honghe deployed a three-node TDengine cluster, each node having a 64-core processor, 256 GB of memory, and 47 TB of storage. Compared with previous systems, TDengine not only supports zero-code OPC UA data ingestion, but also provides more advanced query syntax and computational functions, enabling more effective real-time monitoring and quality analysis of the redrying process. Leveraging TDengine’s high compression ratio, this setup has been able to stably store over 2 trillion time-series records generated by the SCADA system over the long term.

Implementation Results

Honghe’s adoption of TDengine was not merely a database replacement, but a broader upgrade of the technical architecture driven by the core data characteristics of the redrying process. It has helped transform the redrying workshop’s data from a “burden” (high storage costs, slow queries, and difficult analysis) into a “valuable asset” (easy to store, fast to query, and ready for analysis).

By comprehensively addressing performance bottlenecks in data ingestion, storage, and computation for massive time-series workloads, TDengine has made the redrying process more digitalized, transparent, and intelligent. This, in turn, supports key objectives such as stable product quality, optimized process parameters, and reduced operational costs.

Future Plans

Looking ahead, the company is exploring the adoption of TDengine Historian to further enhance its data utilization capabilities. By building on its existing time-series data foundation, the company aims to enable more intuitive data visualization, unified data access, and advanced analytics across its operations. With the integration of historian capabilities, it plans to support real-time monitoring, historical data analysis, and AI-driven applications such as anomaly detection and process optimization. This initiative will further strengthen the company’s digital infrastructure and accelerate its transition toward more intelligent, data-driven redrying operations.