Case Study: Tüpraş achieves up to $20M annual savings with MathWorks (MATLAB) automated control‑loop monitoring

Tüpraş Saves Millions of Dollars Annually with Automated Control Loop Performance Monitoring

Tüpraş, Turkey’s largest industrial company operating four refineries and roughly 5,900 PID control loops, faced a major challenge: monitoring and tuning thousands of control loops manually, a process that consumed hundreds of engineer‑hours and risked production losses and safety issues. After finding packaged third‑party tools costly and hard to customize, Tüpraş turned to MathWorks technology—principally MATLAB along with DSP System Toolbox and Signal Processing Toolbox—to enable an in‑house automated monitoring approach.

Using MathWorks’ MATLAB environment, Tüpraş engineers developed and verified autoregressive models, spectral and cross‑correlation analyses, bicoherence tests, and pattern‑recognition algorithms to create a full control‑loop performance monitoring and root‑cause analysis system. The MathWorks‑based solution, later ported to C# for enterprise deployment, is estimated to save Tüpraş $12–$20 million annually, eliminated about 250 engineer‑days of manual work each year, and avoided the $300,000–$1 million cost of third‑party packages.

Tüpraş

Mehmet Yagci

Modeling and Process Control Superintendent