From Wonderware to TDengine: Modernizing Data Infrastructure at Dali Cigarette Factory

Located in Yunnan Province, one of the world’s premier tobacco-growing regions, Dali Cigarette Factory is a key manufacturing facility under China Tobacco, the world’s largest tobacco company by revenue. China Tobacco produces over 2 trillion cigarettes annually and generates more than $200 billion in revenue, accounting for a significant share of the global tobacco market. As part of this system, Dali Cigarette Factory plays an important role in producing high-quality products for leading brands such as Hongtashan and Yuxi. In recent years, the factory has been actively advancing its digital transformation, adopting industrial IoT, data platforms, and intelligent control technologies to improve product consistency, operational efficiency, and real-time visibility across production.

Background and Challenges

Prior to using TDengine, Dali Cigarette Factory relied on Wonderware due to the need to preserve and utilize historical data. Wonderware’s Historian database is built on Microsoft SQL Server and uses INSQL statements for time-series data storage and querying. However, limitations in its underlying architecture make it increasingly difficult to meet the factory’s requirements for fast data processing.

The large volume of time-series data generated during industrial processes contains significant value, but it is also highly time-sensitive. Collecting, processing, and transforming data in the shortest possible time to unlock the most value had become a primary concern for the factory.

Product Selection

As part of its digital transformation initiatives, Dali Cigarette Factory requires a new data infrastructure solution that could meet its requirements in terms of real-time performance, cost efficiency, and multi-system integration. After extensive evaluation and comparative testing, the factory ultimately selected TDengine for the following reasons:

1. In processes such as tobacco processing and cigarette rolling, key parameters (e.g., moisture content of cut tobacco) fluctuate frequently. TDengine provides second-level real-time control capabilities, significantly reducing the delay in parameter adjustments. In addition, its columnar storage and high compression ratio greatly reduce the cost of storing industrial data.
2. The original Wonderware platform, built on SQL Server, suffered from high latency in real-time data processing and high storage costs. Under the traditional architecture, data from MES, SCADA, and PLC systems was siloed, leading to lengthy root-cause analysis during quality incidents.

Migration Process

In August 2025, the factory deployed a three-node TDengine cluster and used taosX to migrate data from the original system. Applications were then gradually switched over, ensuring a smooth system transition. Thanks to the high performance of taosX, more than 700 billion data records accumulated over three years in the original cluster were successfully migrated, laying a solid foundation for the stable operation of the new system.

Intelligent Process Quality Control Project for Tobacco Processing

During the tobacco processing stage, key processes such as leaf conditioning, flavoring, and drying often face significant fluctuations in process parameters and unstable control performance. To address the difficulty of dynamically optimizing process and equipment parameters, Dali Cigarette Factory launched an AI-driven process optimization system. This system enables real-time optimization of both process and equipment parameters, significantly improving the consistency of cut tobacco quality and overall production efficiency.

The project integrates industrial big data analytics with intelligent optimization and control technologies, deeply combining business operations with a big data platform. In this platform, TDengine acts as the core data infrastructure, extracting, storing, and distributing data from multiple systems — including the tobacco processing control system, cigarette rolling data acquisition system, MES, and others — providing a solid data foundation for process optimization.

At present, the system is able to maintain the product’s outlet moisture deviation consistently within a range that outperforms process standards and previously used manual systems. The table below compares the moisture deviation in Hongtashan and Yuxi cigarettes before and after TDengine was deployed.

	Moisture Deviation – Hongtashan			Moisture Deviation – Yuxi
	Wonderware	TDengine	Improvement	Wonderware	TDengine	Improvement
Loosening	0.49	0.281	74%	0.31	0.244	27%
Casing	0.13	0.109	19%	0.15	0.128	17%
Drying	0.0788	0.0512	54%	0.0772	0.0527	46%

Building on this foundation, Dali Cigarette Factory extended the system architecture of to the leaf threshing and redrying production line, achieving similarly significant results.

Cigarette Rolling and Packaging Process Quality Analysis and Control System

Leveraging TDengine, Dali Cigarette Factory has implemented comprehensive data collection, storage, and analysis for cigarette rolling and packaging equipment, covering more than 40,000 monitoring points. With TDengine enabling centralized ingestion and storage of multi-source data, query efficiency has improved significantly, reducing traceability time from several hours to just minutes.

The system enables real-time monitoring of equipment operating conditions, including key parameters such as speed, temperature, and pressure. When abnormalities are detected (e.g., excessive temperature in a component), the system immediately triggers alerts to notify maintenance personnel for timely intervention.

In addition, through long-term operational data analysis, the system can predict potential failures in advance and support preventive maintenance. This not only extends equipment lifespan but also significantly improves equipment utilization.

Future Plans

Currently, PLC data is written into TDengine TSDB through custom programs developed by system integrators. Whenever new data points are added, the program code must be modified, resulting in higher operational and maintenance complexity. Going forward, Dali Cigarette Factory plans to consolidate all PLC data into a unified OPC UA server and configure taosX ingestion tasks through the TDengine TSDB Explorer web interface. This will enable zero-code data ingestion, significantly reducing the complexity of adding new data points and ongoing maintenance.

Additionally, the factory has begun exploring the deployment of TDengine Historian, especially the migration of Grafana-based visualizations to TDengine IDMP for its AI capabilities and close fit to industrial visualization requirements, enabling enterprise-grade dashboards such as the following.