arthlol / bpswg Goto Github PK

Simulation experiments need to be able to have optional warmup periods to 
arrive at 'steady or desired' state prior to beginning statistics collection.

Parameters for boundary events (Frequency etc)

This "issue" can be used as an index to custom labels associated to (real) 
Issues. Same Issues can be found using Advanced Search.
Unfortunately, can't find a way to display and group by labels in the Issues 
list :(

For additional info, pls contact Razvan:-)

Calendaring time vs time on task references for SLA

Constraints on parameter (availability (time, enum, cost, etc))

On Element parameters:


- Performance parameters: a KPI can be composed by other KPI. (Total sales = 
Total sales aviation business + Total sales non aviation business). The same 
applies to an SLA.

- Cost parameter: Activity cost as a concept (for me) is the sum of all the 
costs (human resources, raw-material, equipment, financial costs). Unless means 
a different thing, Activity cost should be deleted.

- Cost parameter. Equipment is missing (or more broadly asset).

- Payload parameters as my understanding this is the wage of the people that 
execute an given activity? If yes, this cost belongs to human resources – 
above, and should be deleted.

Will our scope include the use of historical execution logs as input for 
process analysis and simulation

The current proposal has scenario as potential inclusions in BPM modeling 
standards. E.g. a BPMN or XPDL including scenario. This leads to a valid BPMN 
or XPDL that has scenario parametrization data included.

Should we revers this architecture to rather  have Scenarios include the 
process model. The resulting files would not be valid BPMN or XPDL files but 
rather PAF scenario files that include a process model (BPMN or XPDL)

Load parameter  (type of load on which the task is carried out)

What are the meaning of the times used to evaluate performance under the time 
dimension?

I’m used to process time – the time the activity takes to be performed, and 
cycle time – the time it takes to the next activity start. I’m ok with set 
up time – used in industrial processes, rework time, but what does it means 
transfer and queue time? And validation? Validation is an activity by itself.

Is there a definition for these?

Cycle time, is this the transfer time?

On Element parameters:


- Cost parameter: Activity cost as a concept (for me) is the sum of all the 
costs (human resources, raw-material, equipment, financial costs). Unless means 
a different thing, Activity cost should be deleted.

Past work item arrivals

Past Task execution time stats (vs various parameters of temporal perspective)

I'm testing the sandboxing using the PAF editor and I would like to understand 
how inserted values are validated. On this short example, I put values on a 
gateway that cannot ever be true (the sum of the branches probability is more 
than 100%).

How the PAF editor leads with this, or better, how he standard is worried with 
data quality and incongruities like this.

Please take a look on the attached files:

- a snapshot 
- full visio file for the ones that installed the paf editor add on.

Can we add a parameter to distributions to let you specify whether a 
distribution is continuous or discrete. This would be helpful in case of say a 
uniform distribution where you need an integer to be returned from a range of 
values. 

Work item type (specific parameters)

Many times in doing analysis users will change only a few of the parameters of 
the scenario.It should be possible to specify these changes without having to 
re specify what does not change in the scenario.

Resource calendaring/availability

Provision of decision logic for routing

Simulation vs Stimulation (of outside systems, interactions)

There is an metaclass called RandomDistribution. I think it is intended to 
represent a uniform distribution, in which case it would be better named 
UniformDistribution. 

We might still want a "RandomDistribution." It would be an abstract superclass 
of all the distributions shown. Going that route, we might also have a 
"NonParametricDistribution."

Scope definition: (Simulation, Analysis, Optimization)

We need to address conformance with respect to both implementation requirements 
and behavior requirements

On Element parameters:

- Cost parameter. Equipment is missing (or more broadly asset).

We already have in our scope to reference elements from both BPMN and XPDL.
Should we also include BPEL in there?

Deadlocks & Bottlenecks

Metamodel and schema for results (output)

Completion sentry 

The model currently provides the relationship between the simulation data and 
the model to which it applies using ElementParameters.references. With some 
coordination with the BPMN 2 spec, and use of UML-style profiling, one could 
achieve this result with stereotyping (by applying a profile to the BPMN).

There are pros and cons to each approach. One drawback of using separate files 
and the ElementParameters.references is the complexity it adds to managing 
files and maintaining proper cross references. 

Our work, of course, cannot be restricted to stereotyping. Most or all of the 
representation of analysis results will require new metaobjects.

Of course, the choice to use stereotype is not to be taken lightly. This is a 
key design issue.

During the BPSWG Kickoff meeting, R Shapiro raised this issue against the 
presented voting system

Providing Examples

We should consider specification of simulation outcomes (results) 

On Element parameters:

- Payload parameters as my understanding this is the wage of the people that 
execute an given activity? If yes, this cost belongs to human resources – 
above, and should be deleted.

There are two potential uses of historical data: 

1. To feed into a curve fitting algorithm as an interim step prior to 
simulation. For example, this may be used to establish a valid baseline that 
the simulation tool is adequately reflecting the historical reality. 

2. To 'replay' exact data with the expectation that the simulation tool will 
produce the same result as happened in reality. 

I think we are agreed that 1 is in scope but that 2 is less clear. 

This issue serves as a reminder to explain clearly what we decide. 

Scope definition: Outcome(questioning the model) vs. Output  

On Element parameters:


- Performance parameters: a KPI can be composed by other KPI. (Total sales = 
Total sales aviation business + Total sales non aviation business). The same 
applies to an SLA.

Resources allocation (prioritization, Selection (Skills, Avail, )

Past Staffing Schedules

If not mistaken, the other standards use xyzRef when referencing other elements 
in the model. Shouldn't we do the same with elementRef (instead of elementId)?
... I hope I'm not saying something that doesn't make sense :-)
Razvan:) 

Some standard efforts have undertaken by WfMC on execution log will we refer or 
make use of it

Need to address resource allocation/availability as input parameters

Typically (e.g. OMG standards) models do not use plurals in naming of 
metaobjects. I think we should adopt the convention of avoiding plurals. 

Currently, names violating the convention include ElementParameters, 
ScenarioParameters, and PerformanceParameters.

This Issue was created by Razvan to test tagging (labeling) capabilities w/in 
Google Project interface.
Please ignore it as an Issue.
Thanks,
Razvan:-)

Goal (Objectives)  Achievement sentry

We should consider specification of process structural (topology) 
variations/assumptions (as input to simulation)

General Usage of Past execution (expectation of structure and content of logs)

Past Branching statistics (Routing)

Add a UserDistribution type to the list of distributions. A UserDistribution is 
made up of empirical data constructed as a list of paired data items. 

Each data pair has a value property (integer or double) and a weight property 
(double). The sequence of pairs forms a probability distribution rather like a 
histogram.

Also a mechanism is required to indicate if the UserDistribution is a discrete 
or continuous distribution (see Issue 47 too) 

Example:

<userDistribution distributionType=”continuous”>
   <entries>
        <entry value="4.5" weight="2.0"/>
        <entry value="5.0" weight="50.0"/>
        <entry value="7.0" weight="80.0"/>
        <entry value="8.9" weight="30.0"/>
   </entries>
</userDistribution>

Integration of Process , Rule, and Events

Resources are defined in BPMN exchange format using the <Resource> element. 
Resources can represent individual people or pieces of equipment. 

Alternatively a group of homogenous resources can be defined e.g. Clerks, this 
usage requires a quantity to be set, which can be changed as the simulation 
runs. 

Resources are used to constrain tasks.