All Member Open Forum

 View Only
Back to discussions
Expand all | Collapse all

Does your plant in some way use the Weibull distribution to manage maintenance?

  • 1.  Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-25-2019 02:54 PM

    Lately, I've been giving a great deal of thought to the contrast of the proportional hazard regression (phreg) model to the Weibull distribution for survival/hazard analysis.

    It has made me wonder. How many plants actually put the Weibull in play? And for the plants that have given it a purpose, what do they use it for?



    ------------------------------
    Richard Lamb
    Analytics4Strategy.com
    Houston TX
    832-710-0755
    ------------------------------


  • 2.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-26-2019 08:34 AM
    Richard,
    Weibull is very good tool for maintenance and reliability.
    You can also use Crow-Amsaa (Weibull Visual) ​, which plots cumulative failures vs. cumulative time, to predict equipment or system reliability.

    ,

    ------------------------------
    Edward Awad
    Gastonia Reliability Engineer
    Owens Corning
    Dallas NC
    ------------------------------

    Original Message


  • 3.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-26-2019 08:50 AM
    Edward,
    What specific routine and ad hoc maintenance and reliability practices does Owens use Weibull for?




    Original Message


  • 4.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-26-2019 10:26 AM
    Richard,

    The short answer is yes. We use Weibull and Growth Analysis in Maintenance and Inspection Strategies.

    We use the following approach or Process as a "Failure Elimination Work Process" :
    1.- Bad actors Identification or Top ten contributors to Production Loss/Maintenance Cost in the last 2 years. We contrast the results with their criticality ,based on that,  we prioritize the following: (For most critical Bad actor to least critical bad actor) (This is a continuous process):
    2.- Collect the Work History (only Failures, Not PM or PdMs) and perform a Growth analysis (Crow AMSAA method) for that particular asset or Group of similar assets (Pump xx1 A/B/C)
    3.- If the system is "SAD" or MTBF decreasing (usually it is the case) then,  we need to do something ASAP, such.. 
       3.1.- Reverse RCA.
       3.2.- and/or Weibull analysis to better understand the Failure Pattern of the predominant  Failure modes.
       3.3 Based on 3.1 and or 3.2 we review the actual strategy and Adjust accordingly. Based on the Beta values of Weibull greater than 1 we can also adjust the PM frequency doing a Cost Risk analysis. 

    Regards



    ------------------------------
    Antonio Alvarez
    Manager
    PwC
    Katy TX
    ------------------------------

    Original Message


  • 5.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-27-2019 09:18 AM
    ​We've used Weibull analysis to determine calibration intervals.  By tracking the As-Found conditions for a particular group of instruments over time, one can determine the appropriate interval given a risk level.  A common risk level is usually 5 to 10% of the population out of tolerance.  The Weibull plot will give the time to point.

    ------------------------------
    Ron Reimer
    Reliability Manager
    Bloomington IN
    ------------------------------

    Original Message


  • 6.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-27-2019 11:38 AM
    Does anyone use Weibull for analytics of issues other than failures? Has the method become associated in our minds with failures instead of time to events, or no events, for a body of cases during a window of time?

    For example, in the full range of questions that can be answered is the optimization of time to return assets to service (https://analytics4strategy.com/tmismnyqstns). At a stage in the maintenance cycle, what is the shape of retention and exit and is it acceptable to us? If not, what will we make it?

    ------------------------------
    Richard Lamb
    Analytics4Strategy.com
    Houston TX
    832-710-0755
    ------------------------------

    Original Message


  • 7.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-29-2019 08:41 AM
    Dear Mr.Richard,
    I have used Weibull analysis for predicting the residual life in electrolytic cell lining material. I am working for a primary Aluminium smelter. We produce Aluminium is produced by electrolysis and the lining of the cells are vital to the plant running at rated efficiencies. The life of the lining was available for different suppliers of the lining material over successive generations,which were subjected to Weibull analysis. The analysis was done on a excel workbook using the Weibull function. The resultant output was used to make reliability estimation over a period of time. This was taken as one of the important inputs for making the annual schedule for cell relining.

    ------------------------------
    Srihari R ramasubramanian.srihari@gmail.com
    ------------------------------

    Original Message


  • 8.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-29-2019 10:49 AM
    There seems to be a common thread. We seem to think and apply Weibull to very focused elements-a material, asset, component-where events are very specific and clear with respect to unambiguous time origin, scale for passage of time and meaning of the failure.

    Does anyone use Weibull for larger picture or sibling methods to deal with larger pictures-e.g., the range of Kaplan-Meir Survivor models. Or step out of those to the Crow-AMSAA model (as Antonio Alverez does)?

    Also, because reliability must also reflect what of the specified did not fail or happen during the window of study, how do studies account for censored cases?

    ------------------------------
    Richard Lamb
    Analytics4Strategy.com
    Houston TX
    832-710-0755
    ------------------------------

    Original Message


  • 9.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-29-2019 08:54 AM
    Dear Mr.Richard,
    The Weibull analysis can give insights on the type of failures that a component is being subject to in the process. By looking at the values of alpha and beta, one can get an idea whether the component has experienced a premature failure or a wear out failure, based on prior operating history. This can enable one to take an informed decision on replacement duration of components based on the MTBF that is experienced by the components on the field. Since this is data that is obtained from the shop floor, it will be a good representative model. This information can be further used for formulating the annual business plan in terms of turnaround requirements,spares and consumables budget, etc.
    I would like to know if anyone is using the Weibull analysis for preventive maintenance. The Weibull analysis seems alright for electronic components which tend to follow the exponential distribution. How about mechanical components? Will the Weibull data be useful in those cases as well?

    ------------------------------
    Srihari R
    ramasubramanian.srihari@gmail.com
    ------------------------------

    Original Message


  • 10.  RE: Does your plant in some way use the Weibull distribution to manage maintenance?

    Posted 03-29-2019 10:51 AM
    In response to Sirihari's comment and question, "I would like to know if anyone is using the Weibull analysis for preventive maintenance. The Weibull analysis seems alright for electronic components which tend to follow the exponential distribution. How about mechanical components? Will the Weibull data be useful in those cases as well?I"

    The answer is yes. As for proactive maintenance, beta greater than one points to a candidate for proactive maintenance. See the examples given by Antonio and Ron.

    ------------------------------
    Richard Lamb
    Analytics4Strategy.com
    Houston TX
    832-710-0755
    ------------------------------

    Original Message