Knowledge-based Engineering: Case Study at Lorain Pipe Mills

Project Objective ] Phase I: Data Collection and Analysis ] Phase II: Data Collection and Analysis ]
 

Lorain Pipe Mills, a division of United States Steel (USS) is located in the west of Cleveland Ohio, in the city of Lorain. With the capability of Lorain Pipe Mills, USS can market pipe in a full size range of 1.9 inches through 26 inches OD. The rotary rolling process developed can produce seamless pipes with lengths exceeding 40 feet. These lengths are necessary to meet the new demands of pipe for oil wells and transmission lines and the high volume, long distance steel pipelines for natural gas.

With the complicated procedure involved in producing very long seamless tubes, there was a problem observed with the Yield of tubing. It was necessary to address this challenge of poor Yield and to model the entire process for better understanding and assist inexperienced users in decision making.

In the month of October 2000, OAI (Ohio Aerospace Institute) together with CAMP Inc. and the University of Cincinnati (Intelligent CAM Systems Laboratory) under the ‘Technology Action Fund (TAF) Project’ decided to assist Lorain Pipe Mills to address the problem cited above. The goal of this project was to develop a prototype system (known as a knowledge agent) to facilitate yield improvement for the seamless rotary tubing process at Lorain Tubular. The project officially commenced in the month of March 2001. After a few sessions of brainstorming, it was decided to focus on the III Seamless Mill which is the only tubular facility in North America capable of producing pipe over 16 inches in diameter. With the complex process of ‘double piercing’, this was the potential site to address the ‘poor Yield’ problem.

After performing the Data collection in the month of October 2001, it was statistically analyzed. The analysis revealed a few problems with the data acquisition procedure and the accuracy of the results was questionable. Hence a second phase of data collection was performed focusing on the all the problems observed during the first phase. This data was carefully analyzed and a knowledge agent was developed. Also, recommendations were made to improve the Yield of tubing. The project was successfully completed in the month of January 2003.

Project Team: Dr. Samuel H. Huang, Mr. Saurabh Dwivedi, and Dr. John J. Shi

Acknowledgement: Thanks to Mr. Shelby Buell and Mr. John Bradley from CAMP, Ms. Dorothy and Mr. Jack from Lorain Tubular for collecting the data.

Table of Contents

PROJECT OBJECTIVE
PHASE I
    1. DATA COLLECTION & PRE-ANALYSIS
        1.1. Summary of the Data Collection
        1.2. Problem of Missing Data
        1.3. How to represent the output (Yield)?
    2. PROTOTYPE YIELD PREDICTION MODEL
    3. STATISTICAL ANALYSIS OF DATA
        3.1. Regression analysis of the data
        3.2. Hypothesis Testing
            3.2.1. t-test for East and West Reelers
            3.2.2. t-test for Plugs A and B
            3.2.3. f-test for Plugs A and B
            3.2.4. Conclusions
PHASE II
    1. PHASE II DATA COLLECTION
    2. ANALYSIS RESULTS
    3. DIMENSIONALLY REDUCED MODELS
    4. EFFECT OF PLUG AND REELER
    5. CONCLUSION
    6.
RECOMMENDATIONS

 

Knowledge Agent for LT

Operation Manual
Software tool to download