Speak to a world-class expert now.
1-800-FOR-KKAI (1-800-367-5524)
info@kkai.com
PrimeTrack® Instant Conferencing
PrimeTrack® Project Management System (patent pending)
CENTER FOR RAPID RESPONSE ENGINEERING® SOLUTIONS
ENGINEERING AND SCIENTIFIC CONSULTING SERVICES
Merging Technologies Solve Rolling Mill Bearing Failures
KKAI Associate
JOP offers expertise in the analysis and design of heavy duty manufacturing machines and equipment; with special interests in the stress, thermal, and dynamic analysis of rolling element bearings and drive trains. He is one of the leading authorities in his field, and the following article illustrates some of his failure anlaysis expertise.
A rash of work-roll bearing failures began when the steel industry converted many rolling-mill stands to work-roll bending systems to make high-quality automotive sheet steel. Bearing failures quickly rose and associated maintenance costs soared to over $1,000,000 per year for some plants.
Bearings also failed on newer mills that were built with heavy bending and roll-shifting systems. Apparently, conventional design methods were no longer adequate for such bearing applications. More frustrating, mill personnel had difficulty correcting the problem, even with the help of bearing engineers.
One major midwest steel company experienced such problems with four-row tapered roller bearings on an 84-in. hot strip mill. These bearings operated trouble-free until roll bending was added on the last four stands of the mill. Then, bearing failures rose from the usual 10 to 14 per year to 42 in the first year and 60 in the second.
After initial investigations into the bearing failures were unsuccessful, company engineers asked JOP to investigate the new bending process and its relevance to the bearing failures.
Roll-Bending Process
A work-roll-bending system applies large bending loads at both ends of the roll, thereby adjusting the profile (bend) of the roll barrel in contact with the steel strip. This profile, in turn, controls the waviness and crown of the strip. Hydraulic cylinders apply 40 to 100-ton loads per side by pushing against the work-roll housing (chock), which transfers the bending load through the bearing to the rotating roll, Figure 1. Prior to roll bending, these bearings normally experienced no radial loads and only residual thrustloads.
Although bearing chocks appear to be rigid structures, they are elastic bodies that can deflect under heavy bending loads. The resultant distortion of the housing bore and bearing cups (outer races) alters the load distribution among the bearing rollers. This can boost stress levels and cut bearing life to a level 40 to 50% of the calculated value for a rigid support based on 40 to 100-ton loads. Therefore, we suspected that chock distortion was causing the problem.
Failure Mechanism
Figure 2 shows the four bearing rows A, B, D, and E and the associated outer races or cups of a failed bearing, The large spalled area in row B was caused by roller edge loading and started at the large diameter end of the cup. A second spalled area (less severe) developed at about 90 degrees from the first on the adjaceut row D. These failure regions are located at about 45 deg clockwise and counterclockwise from the vertical load centerline.
Roller edge loading is usually related to misalignment or bending moment due to offset loading acting on the bearing. On many chocks, the hydrostatic cylinders applied the bending loads away from the axial center of the chock, this creating an offset loading on the bearing. Additionally, there was concern that installation of the work roll via a sled mechanism introduced bearing misalignment.
Initial Analysis
Numerious runs with our COBRA bearing analysis program established the base-line bearing performance as assuming rigid chocks and shaft journals. Then, we superimposed various offset loads and misalignments on the radial and thrust bearing loads to duplicate the bearing pattern. These scenarios all produced adverse bearing conditions. But the roller load patterns and predicted critical bearing regions did not correlate with the mill failures. For example, misalignment and offset loading typically produced critical raceway locations 180 degrees apart, rather than the 90 degrees observed. Aiso, the two outboard cups appeared, at times, to be more prone to failure than the inboard cups for bearings B and D, which failed in service.
Chock Distortion
The next step was to combine finite element analysis (FEA) of the chocks (for both top and bottom rolls) with the bearing analysis. We constructed 3D finite element models of the top and bottom chocks, with simulated bearings. This analysis used a PC version of a commercial program called COSMOS/M. Because of symmetry, only one-fourth of each chock was modeled, Figure 3, thereby reducing model size and run-time.
After calculating chock distortion in the FEA program, we transferred the distortion pattern to the bearing analysis program, then compared the bearing roller load patterns from both programs to assure that the estimated load distribution was very nearly the same.
Figure 4 shows the, roller load distribution as predicted by the bearing and FEA models. This bearing has 46 rollers per row with roller 1 located at top-dead-center (TDC) of the bottom chock. With a rigid chock, roller 1 is the most heavily loaded. When elasticity is added to the bottom chock, the bearing develops two critical regions per row - at 40 degrees clockwise from TDC, which corresponds to roller 6 in Figure 4, and at 40 degrees counterclockwise. These lobes correspond to the spalled regions observed on the failed bearings.
Misalignment was then introduced into the system by rotating the bottom chock about the vertical axis. As misalignment increased, the middle cups (rows B and D) picked up load. Stresses increased in one critical region of row B at the lobe and the second region located 90 degrees away in the lobe of row D. In both locations, roller edge loading was predicted as the source of spall initiation. This spalling would most likely happen at the larger diameter end of the tapered roller bearing cups, as observed in the failed bearings.
Chock Stress
Although the bearing chocks distort in many of these applications, they typically have low stresses and do not fail. For example, the bearing/FEA stress analysis, Figure 5, shows stresses up to only 4,000 psi in the thin sections of the bottom chock. The stress pattern within the bearing circumference shows the double lobed, high-stress regions noted earlier.
Mill Tests
As mentioned, chock distortion can cut bearing life by more than half. Surprisingly, though, the analysis showed that this distortion (with proper alignment) decreased the B10 fatigue life of the four-row bearing set by only 6 to 10%. Therefore, mill tests were conducted to supplement the analytical work and look for additional contributing factors.
To equalize loading and wear, the plant typically moves each bearing/chock assembly to a different location within the four rolling stands that have work roll bending systems. Thus, if a bearing sustained microscopic damage in one location, then it was moved to another location where it subsequently failed, it was impossible to determine if one location imposed more severe conditions than the other.
The plant conducted a series of mill tests in which each bearing/chock assembly was kept in the same location throughout the test. This procedure revealed that the primary and secondary spalling observed in the failed bearings occurred at one particular mill location. The good news was that bearings from the other positions were in good condition. Therefore, attention was focused on the critical mill location.
Solution
Once the contributing factors had been defined, several methods were considered for minimizing or eliminating the misalignment. The easiest fix was to manually realign the work rolls to eliminate misalignment caused by installing the assembly with a sled mechanism. A software patch was added to the process-control computer to ensure the alignment step was taken before resuming production.
The plant then installed a new bearing in the critical location. Recent inspection showed that this bearing has run longer than those that exhibited damage in previous tests. No indications of failure initiation were observed in any of the bearing cups. However, a snaky roller wear path in the critical rows indicated that some misalignment still exists. Additional work is under way to eliminate this remaining misalignment.
To search for specific expertise, enter your search query (type of expertise you are seeking) in the box below, choose to search the entire Rapid-Response site or just resumes and brochures, then click the search button. Please note that the list of resumes available online is a select subset of our vast database. If your search of our site does not reveal the expertise you are seeking, please call us toll-free (in the U.S.) at 1-800-367-5524 or contact us for more information and we will gladly assist you in locating the specific expertise you require.
Have an urgent engineering or scientific need that requires expert testimony, failure analysis or engineering consulting? Contact KKAI now for rapid response from a world-class expert.
1-800-FOR-KKAI (1-800-367-5524)
info@kkai.com
BACK TO THE TOP OF THIS PAGE
Kevin Kennedy & Associates, Inc.
Rapid Response Engineering® Solutions
3905 Vincennes Road, Suite 320
Indianapolis, Indiana 46268
(317) 536-7000 voice
(317) 536-7220 fax