 |
Related Topics: |  |
 |
 |
|
 |
 |
 |
Models are resources that are available for use throughout the project and can be managed via the Resource Manager. Models can be defined manually or they can be created by publishing results from an analysis in a Synthesis-enabled application. For example, you could publish results from a BlockSim diagram or from a Weibull++ life data analysis as a model for use anywhere within the project.
A model that was created by publishing results from an analysis can be viewed in the Synthesis Viewer window, which displays the information about the model but does not allow you to edit it. The Synthesis Viewer window appears when you double-click the model’s name in any location where it is used (e.g., in a URD, in a spare part pool, etc.). It can also be accessed from the Models page of the Resource Manager by selecting the model and choosing Home > Edit > View or by double-clicking the model.
To edit a model that was published from an analysis, you must return to the original data source, make the necessary changes, recalculate and republish the model so it is once again synchronized with the original analysis.
IMPORTANT: Re-publishing a model changes the resource, which is available for use throughout the project. This will cause the model to be changed everywhere that it is used.
You can also manually define a model via the Model window, which can be accessed from the Models page of the Resource Manager by choosing Home > Edit > New, by selecting a model that was manually defined and choosing Home > Edit > View or by double-clicking a model that was manually defined.
There are six categories of models, each serving a specific purpose:
Probability models represent a likelihood of occurrence. They can take three forms:
Reliability models
Probability of failure models
Event occurrence models
These models are used by URDs and by switches in BlockSim.
Duration models represent a length of time. These models are used for task durations and for logistic delays associated with crews and spare part pools.
Cost models can take two forms:
Cost per unit time models are used by tasks, crews and spare part pools for costs that accrue over time.
Cost models are used by tasks, crews and spare part pools for costs that arise on a per incident basis.
In addition, each model can be one of two types:
A constant model represents a fixed probability, duration or cost. For duration models and cost per unit time models, you will also need to enter the time units that the model uses.
Applying a model that uses a fixed probability or duration value to a block causes the block to be considered “static.” A static block can be interpreted either as a block with a reliability value that is known only at a given time (but the block's entire failure distribution is unknown) or as a block with a probability of failure that is constant with time. Systems can contain static blocks, time-dependent blocks or a mixture of the two.
A distribution model represents behavior that varies based on factors such as time and/or applied stress. To define a model that uses a distribution:
You can select a distribution from the drop-down list and then enter the required parameter(s), which will vary depending on the selected distribution.
You can use the Quick Parameter Estimator (QPE), which allows you to estimate the parameters of a distribution based on information you have about the reliability of a product, the probability of an event occurring or the typical duration of a task.
The available distributions and their required inputs are as follows:
Distributions:
Weibull: You can select a 2-parameter or 3-parameter Weibull distribution.
Beta is the shape parameter.
Eta is the scale parameter.
Gamma is the location parameter. This field is available only if you have selected a 3-parameter Weibull distribution.
Mixed: You can select to use a mixed Weibull distribution with 2, 3 or 4 subpopulations. Mixed Weibull distributions are used in cases where multiple subpopulations or failure modes exist within a data set; each subpopulation is represented using a distinct Weibull distribution and the different Weibull distributions are then pieced together to form a continuous function called the mixed Weibull distribution (also known as multimodal Weibull). For these distributions, you will need to specify parameters for each subpopulation. To do this, select the subpopulation in the Subpopulation field, then enter the following values:
Beta is the shape parameter for the selected subpopulation.
Eta is the scale parameter for the selected subpopulation.
Portion is the portion (or percentage) of the total population represented by the selected subpopulation. The sum of the portion values for all subpopulations must be equal to 1.
Notice that the number of the selected subpopulation appears next to each parameter name to indicate the subpopulation the parameter applies to.
Normal
Mean is the location parameter.
Std (standard deviation) is the scale parameter.
Lognormal
Mean is the location parameter.
Std (standard deviation) is the scale parameter.
Exponential: You can select a 1-parameter or 2-parameter exponential distribution. You have the option of defining the exponential parameter as lambda (i.e., failure rate) or mean time (i.e., MTBF). This option can be set on the Calculations page of the Application Setup.
Mean Time is a location parameter.
Lambda is the scale parameter.
Gamma is a location parameter. This field is available only if you have selected a 2-parameter exponential distribution.
G-Gamma
Mu (exp(mu)) is the scale parameter.
Sigma is a shape parameter.
Lambda is a shape parameter. Please note that for any fixed value of lambda, the generalized-gamma distribution is a log-location-scale distribution.
Gamma
Mu is the scale parameter.
K is the shape parameter.
Logistic
Mean is the location parameter.
Std (standard deviation) is the scale parameter.
Loglogistic
Mean is the location parameter.
Std (standard deviation) is the scale parameter.
Gumbel
Mean is the location parameter.
Std (standard deviation) is the scale parameter.
Life-Stress Relationships with Distributions: These models describe behavior based on three variables: reliability, stress and time. They can be used for contained load sharing blocks to describe how the life distribution changes as the load changes. Such models can also be used for standard blocks. For simulation purposes, the use stress serves as an additional block-specific parameter on the model; the block is considered to use only that stress level all the way through the simulation.
Arrhenius-Weibull sets the life-stress relationship to Arrhenius and the distribution to Weibull.
Beta is the Weibull distribution shape parameter.
B is the first Arrhenius parameter.
C is the second Arrhenius parameter.
Use Stress is the stress level at which the model will be used and evaluated.
Arrhenius-Lognormal sets the life-stress relationship to Arrhenius and the distribution to lognormal.
Log-Std is the lognormal distribution scale parameter.
B is the first Arrhenius parameter.
C is the second Arrhenius parameter.
Use Stress is the stress level at which the model will be used and evaluated.
Arrhenius-Exponential sets the life-stress relationship to Arrhenius and the distribution to exponential.
B is the first Arrhenius parameter.
C is the second Arrhenius parameter.
Use Stress is the stress level at which the model will be used and evaluated.
Eyring-Weibull sets the life-stress relationship to Eyring and the distribution to Weibull.
Beta is the Weibull distribution shape parameter.
A is the first Eyring parameter.
B is the second Eyring parameter.
Use Stress is the stress level at which the model will be used and evaluated.
Eyring-Lognormal sets the life-stress relationship to Eyring and the distribution to lognormal.
Log-Std is the lognormal distribution scale parameter.
A is the first Eyring parameter.
B is the second Eyring parameter.
Use Stress is the stress level at which the model will be used and evaluated.
Eyring-Exponential sets the life-stress relationship to Eyring and the distribution to exponential.
A is the first Eyring parameter.
B is the second Eyring parameter.
Use Stress is the stress level at which the model will be used and evaluated.
IPL-Weibull sets the life-stress relationship to the inverse power law and the distribution to Weibull.
Beta is the Weibull distribution shape parameter.
K is the first IPL parameter.
n is the second IPL parameter.
Use Stress is the stress level at which the model will be used and evaluated.
IPL-Lognormal sets the life-stress relationship to the inverse power law and the distribution to lognormal.
Log-Std is the lognormal distribution scale parameter.
K is the first IPL parameter.
n is the second IPL parameter.
Use Stress is the stress level at which the model will be used and evaluated.
IPL-Exponential sets the life-stress relationship to the inverse power law and the distribution to exponential.
K is the first IPL parameter.
n is the second IPL parameter.
Use Stress is the stress level at which the model will be used and evaluated.
© 1992-2013. ReliaSoft Corporation. ALL RIGHTS RESERVED.