Creep Analysis Using Larsen-Miller Parameter Method

[MechEng1995]

I am attempting to utilize the Larsen-Miller Parameter (LMP) Method to determine the remaining life (Creep failure) of superheater and reheater sections of coal fired boiler (1.1M lbs/Hr, 1850psi /1000F/1000F). Upon application of this method I am realizing that there are many different opinions on this method and what constant (C) to use. I have also been trying to locate LMP curves and constants for the common boiler tube materials as well as tips on getting accurate results with this method.

Can anyone offer any help or tips?

Thanks.

 

[jte]

I would take the values as given in API 530. See parts 5.6 and H.3. Another source with the same data is API 579.

jt

 

[davefitz]

There will remain a couple of uncertainties that will effect the accuracy of the result.

First, the published larson miller curves are based on the average tested properties or samples +_ 2 standard deviations , to represent a "worst "case. The factors that had led to the variations in the original samples will also exist in any other sample, based on variations in manufacturing techniques and trace elements.

Second, if you are trying to predict the remaining life of the heated tubes, you will need to determine the increase in metal temperature in the heated zone due to heat transfer vs the monitored temperature in the unheated zone. There are significant variations in tube to tube metal temperatures due to uneven flow of fluid inside the tubes and also uneven flow of gas over the tube, and uneven ash and slag coverage.

Third, there are recent rumblings in teh academic and research community that the constant of proprotionality used in Larson Miller plots is not correct.

 

[jte]

Well, in both 530 and 579 the averages are published as well as the minimums. So the worst case is based on minimum; the average is a bit higher.

Certainly there is a fair amount of uncertainty in failure prediction. As davefitz noted, the local flow and heat transfer issues play a role as well as material properties and the specific history of the tubing at that particular location. No matter how good your archived plant operating data is, I'd be willing to bet that they do not have detailed data for a particular 6" section of a particular tube. So to a large extent, these analyses are only slightly better than a SWAG. But it beats a WAG.

jt

 

[metengr]

The Larson Miller (LM) parameter is not really desirable for long range prediction of creep rupture properties in service aged steels. This technique is better suited for comparison of heats of steels or extrapolation of short term creep rupture properties (below 100,000 hrs) under anticipated service conditions. When the LM parameter was originally developed and published years ago, others began to evaluate the relationship of creep properties out beyond 100,000 hours using accelerated testing. It was discovered that other relationships are better suited to fit the creep data obtained from accelerated testing, such as the Monkman-Grant method.

What I have found and reviewed in technical literature is that the better approach for extrapolation of long term creep rupture properties from short term lab data is to use the Monkman-Grant method.

Having tested many a boiler tube, I have found that the remaining creep life is highly dependent on location in the boiler and remaining wall thickness – which relates to service stress. Yes, temperature is important but for all practical purposes, local stresses in the tubes seem to drive the failure mechanism.

There are many operating variables that effect the long term creep properties in boiler tubes – what I believe you can obtain from creep testing is a reasonable estimate of operating time when you need to decide on replacement before you begin to fall off the edge of the cliff with creep rupture failures. The key is tube sampling locations in the boiler.