Evgeny,

I do not recall having an Olympic game at home. Even for a country with hundreds of thousands jobs per year above a million dollar level Olympic games do happens at minimum every 25 years so you cannot say planning and execution of an Olympic game is representative of scheduling practice. 

Still there are a few contradictory statements in PM documents, old concepts might still be influencing their vision. In general I am in agreement with PMI, they are a step ahead of many others. The major issue the PMI have is with what their certification mean. I agree with them that there shall be different certification levels and that an introductory level is a good thing. 

Going back to the theme of weather modeling using statistical methods, if using 3 scenarios approach you model 3 deterministic schedules and there you can apply 3 weather calendars one for optimistic, one for probable and one for pessimistic schedule. Under this method modeling for weather is easy and required computer time is a small fraction compared to Monte Carlo. Computer time can be a practical limitation under big portfolios of hundreds of jobs each with thousands of activities, here 3 scenarios approach shines as a practical alternative. 

Monte Carlo approach is different approach, here you would be modeling hundreds or thousands of iterations, each with a random rain calendar on every scenario run, and this is what I am trying to understand with regard to weather models. You are not to use Monte Carlo and model the same weather pattern on every iteration, every year each specific month rainfall shall be different, not every year is a rain year, not every year is a drought year, it can be a very complex model while the 3 scenarios approach is more straightforward especially if you use the reduced hour approach for your calendars.  At the moment it seems like application of such calendar on Monte Carlo can be wrong unless I am missing something and before running each iteration the model randomly selects a different rain scenario for each and every single iteration. At the moment my Monte Carlo approach is based on modeling random weather/rain events that will impact/delay weather sensitive activities through a weather resource, and this requires the Monte Carlo model to apply resource leveling on each iteration. 

If you run a single iteration Monte Carlo simulation and a multiple iterations Monte Carlo Run fixing all variables, including activity durations, except weather then the results shall differ, otherwise you are modeling the same rain every year, falling under the flaw of averages. 

But I do not use P6 so it will be impossible for me to investigate in depth how their weather models work. 

Best Regards,

Rafael

Evgeny,

This analysis I do on my own as the CPM culture at home and suppose in the whole US is way back into the dark ages. It is rarely required and therefor not many have appreciation for it. 

Most of our specifications are against using transparent buffers, we must hide it within activity durations and therefore any benefit on using statistical approach is useless. We are required to use in our schedules all available time so the idea on targeting for earlier schedule and use of buffers in the hope of meeting contractual dates is forbidden. So much that if in order to increase probabilities of success you submit an early finish schedule it will be considered as a reduction in contract time. There is no purpose on finding probabilities of success when you cannot take action to make your plan to have higher probabilities of success because of claim paranoia.

Most of our schedulers are engineers so the reason is not because of the technical capacity as all engineers have some training into statistical methods. Statistical methods are used everyday in quality control procedures almost every engineer at some point must deal with. 

At the moment I am taking my spare time to learn better scheduling methods. I am doing it using simple models you can easily track. As a matter of fact I believe there is a possibility some of the actually used methods might be wrong and just the fact that weather models predict longer durations does not imply they are correctly used. 

Just by looking at federal government specs you will notice how little these methods are understood. Most of these specs require you to use NOAA data to determine probable rain days but none tells you what is the definition of a rain delay day. It is usually defined as a day with over certain amount of rain per day, but somehow the specs miss the basic definition. To me it is an indication not serious attention is being paid to it and that many specifiers do not even understand what they are asking for. A lot of Bull-Sh..-ting.

The following is an example of such spec but I was ruled out of bidding because of government common practice to specify brand name software.

http://www.mediafire.com/view/yleucq6r9j9903d/Specifications%20-%20VA%20...

• Within 30 calendar days (45 calendar days on projects over $50,000,000) after receipt of Notice to Proceed, the Contractor shall submit for the Contracting Officer's review; two copies of the complete “Day 1” network diagram on sheets of paper 765 x 1070 mm (30 x 42 inches) and an electronic file in a compressed Primavera (P6), (PDM) format.
• The final network diagram in its original form shall reflect original contract scope of work and contain no contract changes or delays which may have been incurred during the final network diagram development period and shall reflect the entire contract duration as defined in the bid documents.
• Schedule Risk Analysis – The contractor shall conduct the statistical schedule risk analysis based on the above detailed construction activities in the Day 1 approved diagram, identifying major schedule risk areas and recommended risk mitigation plans as outlined below.
  • But those who write the specifications do not understand that best risk mitigation strategy is to target for early completion schedule.

Best Regards,

Rafael

Vladimir,

I figured out that by showing the fields on the middle pane I would be able to scroll it easy next to activity description on left pane so the visual issue was never there. It is more a matter of making it easier to select the fields.

I wonder if the custom group is to be implemented if it can be saved on a separate file for export/import, then a user can have unlimited custom field groups.

I am testing ways to model weather using Spider Monte Carlo and up to now everything is going well except when starting to play with opt-exp-pes values for lag. I want to see the obtained distribution for lag, easy for everything else, a piece of cake for every activity duration. For the moment I am considering to create a hammock to make it visible. Any suggestions here will be welcomed.

For some parameters the opt-exp-pes values are protected by some data entry validation while for others it looks like the validation is missing. Wrong values do not damage the model, just hold MC until fixed, the advantage of data validation during data entry is that it prevents the occurrence of many errors that might be difficult to find. 

I love the way Spider Monte Carlo runs, I love the idea of making it possible to model opt-exp-pes lag values, something missing in all/most other software. Long ago I got an appreciation for the 3 scenarios method, but this is a decades long proved method within Spider. 

Best Regards,

Rafael

What about adding an option to select all fields that can have Opt-Exp-Pes values, in addition to actual groups of fields in an additional Group of fields so it will make it easier for the user to identify and select those fields.

Maybe a Custom Group can be also added where users can be able to populate the group as they wish. 

It is possible the kitchen door is still open. 

Best Regards,

Rafael

Perhaps a new Tab shall be created as to include here all Monte Carlo Parameters such as:

PARAMETER -------------------------------------OPT VALUE-----------EXP VALUE------------PES VALUE

REMAINING DURATION(DAYS)

REMAINING DURATION(HOURS)

TOTAL DURATION OF TEAMS(DAYS)

TOTAL DURATION OF TEAMS(HOURS)

CALENDAR

VOLUME REMAINING

PRODUCTIVITY

QUANTITY

WORK LOAD

COST

any other