Wednesday, February 4, 2009

Use static and dynamic results to validate process models

 
IBM Certification Test 992.4 - Analysis and Validation
 

How to Use static and dynamic results to validate process models

You can use static and dynamic results to validate process models.
 
  1. Use static analysis to validate your model prior to simulation.
  2. Use dynamic analysis to validate your model prior to implementation.

Analyzing static process models (static analysis)

WebSphere® Business Modeler enables you to perform various types of analysis on your models.

 
WebSphere Business Modeler provides a variety of static analysis functions that let you extract specific types of important business information from your models. Static analysis can provide relative costs for multiple scenarios, but, most importantly, it provides a means of validating your model prior to simulation

 

Analyzing process simulation results (dynamic analysis)

Analyze the results of process simulation runs to extract specific information on the monetary results, time results, and statistics generated from the running of a process simulation.
 
You must create a simulation profile and run a simulation before you can do an analysis of the simulation results.
 
You can analyze the data from a simulation run at three levels of granularity:
  1. Aggregated analysis
    Aggregated analyses are the most broadly scoped of the dynamic analyses. They use all the data from the entire simulation run for their information. Use these analyses to gain an understanding of the behavior of the process as a whole.
  2. Process cases analysis
    Process cases analyses use the data from specific process cases to give information. Use these analyses to gain an understanding of a specific case (process flow) within the process.
  3. Process instance analysis
    Process instance analyses are the most granular of the dynamic analyses. They use data from a single instance of a process case for their information. Use these analyses to gain an understanding of the behavior of a specific token's run through the process.

Analyzing aggregated results
Perform aggregated analyses to gain an understanding of the behavior of the simulation run as a whole. Aggregated analyses are the most broadly scoped of the dynamic analyses. They use all the data from the entire simulation run for their information.

You need to run a simulation based on a process that you have modeled before you can do an aggregated analysis.

 

Related tasks

 

Analyzing process cases
You can perform several types of process case analysis to determine the performance of a simulated process across all of its process instances. Process case analysis provides statistical information that takes into account all process cases and their probabilities of occurrence.

You need to run a simulation based on a process that you have modeled before you can do a process case analysis.

A process case is a form that a process instance can take, in which units of work follow a particular pattern of connections and activities within a branching process flow. Process case analyses show weighted average values for the results of simulated processes, where the average value is weighted for each process case to account for the distribution of process instances to that case relative to other cases.

You can use process cases analysis when you want to examine process-level data that is gathered from all process instances in a simulation, in order to understand how a process performs when it is run a large number of times. Process cases analysis organizes and averages the simulation results generated for each process case, which enables you to investigate variations in performance between different patterns of process flow. As a result of these investigations, you may determine that you need to modify a process model to improve the overall results or the results for one or more cases. In addition to results for each case, process cases analysis also provides weighted average results that take all cases into account.

By comparison, aggregated analysis enables you to examine results for specific elements within a process (aggregated across all process instances in a simulation without consideration of case matching), and process instance analysis enables you to examine results within a specific process instance. Process comparison analysis is related to process cases analysis in that it compares the weighted average results for two distinct simulation runs.

 

Related tasks

 

Analyzing process instances

Perform process instance analysis to get a detailed understanding of each individual pass through a process during simulation.

Before you carry out a process instance analysis, you must have a set of simulation results that you want to analyze. The first process instance analysis you must carry out is a process instance summary analysis. You can then access the additional process instance analyses from the results page of the summary.

During simulation, every pass through the process creates a process instance, and the data from that pass is recorded for analysis. Use process instance analysis when you want to study the results of a specific process instance. The value of doing this is that it enables you to examine how the process works on a process run level. For example, in simulation of the handling of calls at a call center, you can examine the call with the best or the worst results for elapsed duration or total cost to determine the factors that contributed to those results and to identify potential improvements.

Process instance analysis is the most granular of analysis types and can be used to gain an understanding of how each run through the process occurred during simulation.

 

Related tasks

 

Perform one of the several comparative analyses to find differences and similarities between the results of two different simulation runs.
 
Perform one of the several comparative analyses to find differences and similarities between the results of two different simulation runs.
 
Before performing this analysis you must have two simulation results that you want to compare. For the following comparisons, which are intended to highlight differences in how a process performs when simulation settings are changes, both results must come from the same simulation snapshot:
 
  • Processes Resources Time Comparison
  • Processes Resources Cost Comparison
  • Processes Classifier Duration Comparison
  • Processes Classifier Cost Comparison
 
The other comparisons can be carried out against any two results in the same project.
 
Use processes comparison analysis to highlight differences in simulation results between versions of a modeled process, simulations of different sizes, or among simulations with different decision probabilities. For example, you can use processes comparison analysis to compare the average length of time it takes to complete your as-isprocess versus your proposed to-be processes. When you make changes to a process to optimize cost or throughput performance, you may want to perform several types of comparison analysis to confirm that your intended optimizations are realized, and that there are no unwanted side effects.
 
The average values presented by the comparison analysis are identical to those presented by the process cases analyses.

 

Related tasks
 

 

  • Analyzing activity durations
    The activity duration analysis shows the average time it takes for each activity in a simulation to complete. It also indicates how much of that time is taken waiting for resources to be available.

    You can use this analysis to determine which activities in a process require the longest durations to complete. After identifying these activities, you may then decide to streamline these activities, or to add more resources to reduce delay durations and elapsed durations.
  • Analyzing process cases summaries
    Perform this analysis to display summary information for each of the process cases produced during a simulation.

    Use this analysis when you want an overview of the process cases generated in a simulation. This analysis provides high level summary information for each process case, including duration and cost information and indicates whether the process case reached a successful completion or not. It lists the activities that make up each case and, when possible, quantifies their average cost and average elapsed duration. By selecting a process case in the analysis view, you can highlight the paths of that process cases in the Simulation view.

    This analysis, like other process case analyses, may reveal unexpected results within specific process cases. For example, you may determine that the average duration in a particular process case is unacceptably high. As a result of reviewing the information that this analysis presents, you may decide that you need to modify a process model or reset resource levels, or you may determine that you want to investigate further with another type of process case analysis or process instance analysis. Alternatively, you may investigate the reasons that cause a particular process case to fail.
Related reference

Process cases summary analysis

To do a comparison analysis, complete the following steps:
  1. In the Project Tree view, right-click the first simulation result that you want to analyze, and then select Dynamic Analysis > Processes Comparison Analysis and then one of the following choices:
    • Processes Duration Comparison
    • Processes Activities Total Time Comparison
    • Processes Cost Comparison
    • Processes NPV / IRR Comparison
    • Processes Break Even Comparison
    • Processes Resources Time Comparison
    • Processes Resources Cost Comparison
    • Processes Classifier Duration Comparison
    • Processes Classifier Cost Comparison
    A window opens.
  2. Select the second simulation results that you want to compare with the first results you selected and then click OK.
  3. Select the type of process instances you want to include in the analysis:
    • Succeeded process instances only
    • Failed process instances only
    • All process instances
  4. If you selected the processes NPV / IRR comparison, complete the following steps in the window that appears:
    1. Specify the Initial cost of the process, and optionally select a currency.
    2. In the Number of process runs per year field, specify the number of times you expect the process to run within a single year. For example, enter 5000 to specify that the calculation will be based on an assumption that the process will running five thousand times each year.
    3. In the Number of years field, specify the number of years you want the analysis to cover. For example, enter 2 to determine results based on two years of operation as the payback period.
    4. Enter the Annual discount rate. For example, enter 10 to specify a discount rate of ten percent.
    5. Click Finish.
  5. If you selected the processes break even comparison, complete the following steps in the window that appears:
    1. In the Process Volume field, enter the number of times you want to use as the estimated number of times the process will run in a given unit of time. Then, select the unit of time to use from the Time unit drop-down list. For example, if you want to estimate that the process will run 5 times each hour, select Process volume to 5 and Time unit to hour.
    2. In the Initial cost field, enter the fixed cost value that you want to recover and select a currency type from the drop-down list.
    3. Click Finish.
Now, you can view the results of the analysis in the Analysis view. You can also print a report of the results.

 

Related links


No comments: