## Tips on using this forum..

**(1) **Explain your problem, don't simply post "This isn't working". What were you doing when you faced the problem? What have you tried to resolve - did you look for a solution using "Search" ? Has it happened just once or several times?

**(2)** It's also good to get feedback when a solution is found, return to the original post to explain how it was resolved so that more people can also use the results.

# SDPM

Sun, 2010-02-14 01:15

I want to learn SDPM. I wanted to learn Critical Chain (And I’m reading Mr. Eliyahu M. Goldrat’s books) But then I found Spider Project, based on SDPM rather than CPM.

The real question is:

Can I learn SDPM without using Spider?

How?

Sun, 2010-07-18 11:16

#1
Rafael,

if to follow PMBOK Guide risk management process quantitative risk analysis follows qualitative risk analyisis.

At qualitative risk analysis stage we select those risks that shall be considered and modelled.

Usually risks that have very low probability are not selected for simulation but may be selected to be considered at technical design.

You know that at different areas we have different probabilities and expected strengths of earthquakes.

So local requirements for building design are different at different areas. But it does not mean that larger eartquake will never happen.

Usually it is necessary to consider those project risks that have reasonable probability to happen during construction.

But the number of risks that shall be considered in design is larger because the life cycle of the building (or anything) is longer.

In your example the stadium should be designed considering waterspout danger if such event is not unique but in the construction contract such event shall be interpreted as force majeure.

if to follow PMBOK Guide risk management process quantitative risk analysis follows qualitative risk analyisis.

At qualitative risk analysis stage we select those risks that shall be considered and modelled.

Usually risks that have very low probability are not selected for simulation but may be selected to be considered at technical design.

You know that at different areas we have different probabilities and expected strengths of earthquakes.

So local requirements for building design are different at different areas. But it does not mean that larger eartquake will never happen.

Usually it is necessary to consider those project risks that have reasonable probability to happen during construction.

But the number of risks that shall be considered in design is larger because the life cycle of the building (or anything) is longer.

In your example the stadium should be designed considering waterspout danger if such event is not unique but in the construction contract such event shall be interpreted as force majeure.

Sun, 2010-07-18 09:19

#2
Sat, 2010-03-27 14:29

#3
About the application of SPDM:

http://www.pdfone.com/download/7_keyword-spider-project/the-application-...

Also the following can be of use for those new to Spider Project.

http://www.pdfone.com/download/0_keyword-spider-project/spider-project.pdf

http://www.pdfone.com/download/7_keyword-spider-project/the-application-...

Also the following can be of use for those new to Spider Project.

http://www.pdfone.com/download/0_keyword-spider-project/spider-project.pdf

Tue, 2010-03-02 08:39

#4
The above chart is posted at the following link for your download of the Excel file. It can provide you with a view on how the curves are determined using observed data.

http://rapidshare.com/files/357841934/Distribution_Curves.xlsx

note: first cumm. value should be 102 instead of 2 you can change it yourself, sorry for the inconvenience

Sun, 2010-02-28 12:11

#5
Vladimir,

Of course the ratios remain equal.

Mike,

This triangular distribution is a simplification of what Spider does; it might be using a curve fitting that makes more sense, this because it is based on three schedule estimates and not in hundred or thousand iterations as in Monte Carlo. Remember if you do not have an individual distribution per activity there is no case in going into such granularity. The curve estimate is substantially better than the erroneous original PERT computations based on a single model and asuming an unchanged critical path.

Carlos is looking for a simplified conceptual understanding and he is doing the correct exercise, same we did with a manual forward and backward pass when learning the CPM basics, thereafter we leave the computer to do it for us.

For your knowledge I know nothing about the algorithms Spider uses, basically just a conceptual knowledge, that is all we need and that is what Vladimir is providing.

Best regards,

Rafael

PS to Mike - My older brother Miguel/"Mike" lives in Santiago de Chile, he had quite an experience with the earthquake. He is a devil, makes you pale in comparison.

Photos posted at a Chilean newspaper site

http://www.fotos.emol.com/index.asp?G_ID=13635

Tsunami in BioBio

http://www.fotos.emol.com/index.asp?G_ID=13647#

Tsunami in Pelluhue

http://www.fotos.emol.com/index.asp?G_ID=13648#

Of course the ratios remain equal.

Mike,

This triangular distribution is a simplification of what Spider does; it might be using a curve fitting that makes more sense, this because it is based on three schedule estimates and not in hundred or thousand iterations as in Monte Carlo. Remember if you do not have an individual distribution per activity there is no case in going into such granularity. The curve estimate is substantially better than the erroneous original PERT computations based on a single model and asuming an unchanged critical path.

Carlos is looking for a simplified conceptual understanding and he is doing the correct exercise, same we did with a manual forward and backward pass when learning the CPM basics, thereafter we leave the computer to do it for us.

For your knowledge I know nothing about the algorithms Spider uses, basically just a conceptual knowledge, that is all we need and that is what Vladimir is providing.

Best regards,

Rafael

PS to Mike - My older brother Miguel/"Mike" lives in Santiago de Chile, he had quite an experience with the earthquake. He is a devil, makes you pale in comparison.

Photos posted at a Chilean newspaper site

http://www.fotos.emol.com/index.asp?G_ID=13635

Tsunami in BioBio

http://www.fotos.emol.com/index.asp?G_ID=13647#

Tsunami in Pelluhue

http://www.fotos.emol.com/index.asp?G_ID=13648#

Sun, 2010-02-28 11:47

#6
Mike,

we discuss manual computation of probability to be on time based on finish dates of optimistic, most probable, and pessimistic scenarios analysis.

I am sorry for hurting your brain.

Best Regards,

Vladimir

we discuss manual computation of probability to be on time based on finish dates of optimistic, most probable, and pessimistic scenarios analysis.

I am sorry for hurting your brain.

Best Regards,

Vladimir

Sun, 2010-02-28 11:42

#7
Rafael,

it is not necessary to make big triangle area equal to 1 if to discuss area ratios, not values.

Best Regards,

Vladimir

it is not necessary to make big triangle area equal to 1 if to discuss area ratios, not values.

Best Regards,

Vladimir

Sun, 2010-02-28 11:41

#8
Hi Rafael

More like concussion!

I have no idea what you lot are going on about.

My brain hurts.

Best regards

Mike Testro

More like concussion!

I have no idea what you lot are going on about.

My brain hurts.

Best regards

Mike Testro

Sun, 2010-02-28 11:33

#9
Carlos,

The surface of the big triangle is indeed 1. By dividing graphical area as to make big triangel equal to one we get the relative areas of the other figures. But yes, Agustin is correct, it was a math trick I did without being clear with the intermediate steps. Hope this did not created any confussion.

Best regards,

Rafael

The surface of the big triangle is indeed 1. By dividing graphical area as to make big triangel equal to one we get the relative areas of the other figures. But yes, Agustin is correct, it was a math trick I did without being clear with the intermediate steps. Hope this did not created any confussion.

Best regards,

Rafael

Sun, 2010-02-28 11:18

#10
I missed again something in showing the images.

Augustin

Augustin

Sun, 2010-02-28 11:17

#11
Hi Carlos,

I’ll try to give you a small example of manual calculation following Rafael and Vladimir explanations and showing some images.

Let’s consider a small project consisted of 10 activities. After the risk identification, risk qualitative analysis and risk response planning, there will be different activity durations and consequently, different project durations in the three scenarios. Please take into consideration the fact that most probable and pessimistic scenarios may have much more activities than the optimistic scenario because of the risk events and uncertainties. But for the sample, we can consider all the scenarios having the same list of activities.

<*img src=“http://s833.photobucket.com/albums/zz253/augustin_purnus/?action=view¤t=SDPM1.jpg“ WIDTH=600 HEIGHT=800 BORDER=0/*>

If we apply CPM computation, we will get the following project durations: Optimistic Scenario = 20 days, Most Probable Scenario = 28 days and Pessimistic Scenario = 43 days.

The Triangular Distribution will be limited by the optimistic and pessimistic duration. The interpretation will be: the probability to finish the project at the optimistic duration is 0%, and the probability to finish the project at the pessimistic duration is 100%.

<*img src=“http://s833.photobucket.com/albums/zz253/augustin_purnus/?action=view¤t=SDPM2.jpg“ WIDTH=600 HEIGHT=400 BORDER=0/*>

The surface of the triangle is equal to one.

Now, you may have two situations:

1. You may want to finish the project with a reasonable probability,

or

2. You have a contractual finish date and you would like to know what is your probability to achieve that date

Hope the images can be seen …

In the next message I will include some formulas for both situations.

Best regards,

Augustin

I’ll try to give you a small example of manual calculation following Rafael and Vladimir explanations and showing some images.

Let’s consider a small project consisted of 10 activities. After the risk identification, risk qualitative analysis and risk response planning, there will be different activity durations and consequently, different project durations in the three scenarios. Please take into consideration the fact that most probable and pessimistic scenarios may have much more activities than the optimistic scenario because of the risk events and uncertainties. But for the sample, we can consider all the scenarios having the same list of activities.

<*img src=“http://s833.photobucket.com/albums/zz253/augustin_purnus/?action=view¤t=SDPM1.jpg“ WIDTH=600 HEIGHT=800 BORDER=0/*>

If we apply CPM computation, we will get the following project durations: Optimistic Scenario = 20 days, Most Probable Scenario = 28 days and Pessimistic Scenario = 43 days.

The Triangular Distribution will be limited by the optimistic and pessimistic duration. The interpretation will be: the probability to finish the project at the optimistic duration is 0%, and the probability to finish the project at the pessimistic duration is 100%.

<*img src=“http://s833.photobucket.com/albums/zz253/augustin_purnus/?action=view¤t=SDPM2.jpg“ WIDTH=600 HEIGHT=400 BORDER=0/*>

The surface of the triangle is equal to one.

Now, you may have two situations:

1. You may want to finish the project with a reasonable probability,

or

2. You have a contractual finish date and you would like to know what is your probability to achieve that date

Hope the images can be seen …

In the next message I will include some formulas for both situations.

Best regards,

Augustin

Wed, 2010-02-24 06:41

#12
Carlos,

About the formula:

Capital “S” = area of big triangle = area under probability curve = 1.00 a basic probability rule

Lower “s” = area under small triangle to the right

1.00 – s = cumulative probability at T (target value)

The following is a graphical computation of the cumulative probability at T using adobe acrobat to graphically find the areas of the big triangle and area of the figure to the left of the target.

Cumulative Probability = 16.48/20.16 = 0.82 ~ 0.80

It is just an approximation using a screen capture and adobe acrobat, selecting points using the mouse after a few (14ea) Coronas.

Best regards,

Rafael

About the formula:

Capital “S” = area of big triangle = area under probability curve = 1.00 a basic probability rule

Lower “s” = area under small triangle to the right

1.00 – s = cumulative probability at T (target value)

The following is a graphical computation of the cumulative probability at T using adobe acrobat to graphically find the areas of the big triangle and area of the figure to the left of the target.

Cumulative Probability = 16.48/20.16 = 0.82 ~ 0.80

It is just an approximation using a screen capture and adobe acrobat, selecting points using the mouse after a few (14ea) Coronas.

Best regards,

Rafael

Wed, 2010-02-24 04:57

#13
Hi Carlos,

I opened your schedule.

If activities are not linked Spider Project decides that all activities can be done in every order. If some schedule cannot be improved (shortened) it remains as it is. So Spider tries to optimize the schedule but if the order does not matter you will get some solution.

If you want certain order you shall add links or priorities.

The schedule can be sorted by your criteria.

For an example if you will sort by column Start activity order in each phase will be shown in the order they are executed.

If you will select Detailed Information in Levelling dialog all interruptions of work will be shown in the Gantt Chart.

Best Regards,

Vladimir

I opened your schedule.

If activities are not linked Spider Project decides that all activities can be done in every order. If some schedule cannot be improved (shortened) it remains as it is. So Spider tries to optimize the schedule but if the order does not matter you will get some solution.

If you want certain order you shall add links or priorities.

The schedule can be sorted by your criteria.

For an example if you will sort by column Start activity order in each phase will be shown in the order they are executed.

If you will select Detailed Information in Levelling dialog all interruptions of work will be shown in the Gantt Chart.

Best Regards,

Vladimir

Wed, 2010-02-24 03:21

#14
Carlos,

the formula shows the ratio of the triangle area to the right from the blue line to the area of the whole triangle.

There is some problem with your text files. Please save your project as Spider file - I will look at your example.

If you assign two resources in one team then these resources will work only together. So if your calendar is one hour in a day then others in your team will work also only one hour.

If you want to help only one hour in a day and let others work by their own calendars you shall assign new team, and then yourself at this new team. In this case everybody will work at their own time. Don’t forget to define your productivity.

You shall create three project versions basing on your reference-books. Applying optimistic reference-book to your project you will get optimistic version that shall be saved with one name, applying most probable reference-book you will get most probable version that shall be sent with another name, and applying pessimistic reference-book you will get pessimistic version that shall be saved with the third name. In the risk analysis dialog you shall select optimistic, most probable, and pessimistic project versions (not reference-books) and then run risk analysis.

Don’t forget to look at First Project section of the Spider Help - there are instructions based on the example that is included in Spider Demo.

Best Regards,

Vladimir

the formula shows the ratio of the triangle area to the right from the blue line to the area of the whole triangle.

There is some problem with your text files. Please save your project as Spider file - I will look at your example.

If you assign two resources in one team then these resources will work only together. So if your calendar is one hour in a day then others in your team will work also only one hour.

If you want to help only one hour in a day and let others work by their own calendars you shall assign new team, and then yourself at this new team. In this case everybody will work at their own time. Don’t forget to define your productivity.

You shall create three project versions basing on your reference-books. Applying optimistic reference-book to your project you will get optimistic version that shall be saved with one name, applying most probable reference-book you will get most probable version that shall be sent with another name, and applying pessimistic reference-book you will get pessimistic version that shall be saved with the third name. In the risk analysis dialog you shall select optimistic, most probable, and pessimistic project versions (not reference-books) and then run risk analysis.

Don’t forget to look at First Project section of the Spider Help - there are instructions based on the example that is included in Spider Demo.

Best Regards,

Vladimir

Wed, 2010-02-24 01:14

#15
Thanks, I’m still trying to understand the logic behind the

"1–X = [(P-T)*(P-T)]/[(P-M)*(P-O)]"

formula

I understand that, if reducing success probability, the project will be under more risk (so we can offer the client less time, less costs, more risk), but increasing probability will also increase time and costs. It is just I’m not getting the relationship between the model and the formula. But don’t tell me, I must end figuring it by myself.

That’s one topic, the other is the schedule I am making.

You can download the database here:

http://dl.dropbox.com/u/245847/Translation.zip

I did not include links between activities. I can have activity 16 without having activity 15, if I had another translator I could assign some activities to that other person and cut the deliver time... Instead, I assigned myself as a resource and added a calendar of 1 daily hour of work. Ten scheduled the resource critical path.

At first, in the Activity gantt it looks like a mess. When you look at the Resource Gantt it makes more sense, so my first question in this is:

How did Spider decide which activities are first and wich are after?

The other thing I did was creating reference books for optimistic, most probable and pessimistic productivities. But when I tried to run the Risk Analysis, it gave me an error:

http://screenr.com/pwx

I don’t know how to correct that error. Any help?

I got some stress as I said that I would be delivering the schedule tonight but ... well, my purpose with this project is precissely making errors, and it is taking me time to understand how SDPM is calculated and how I am supposed to create a SDPM schedule.

Best regards,

Carlos.

"1–X = [(P-T)*(P-T)]/[(P-M)*(P-O)]"

formula

I understand that, if reducing success probability, the project will be under more risk (so we can offer the client less time, less costs, more risk), but increasing probability will also increase time and costs. It is just I’m not getting the relationship between the model and the formula. But don’t tell me, I must end figuring it by myself.

That’s one topic, the other is the schedule I am making.

You can download the database here:

http://dl.dropbox.com/u/245847/Translation.zip

I did not include links between activities. I can have activity 16 without having activity 15, if I had another translator I could assign some activities to that other person and cut the deliver time... Instead, I assigned myself as a resource and added a calendar of 1 daily hour of work. Ten scheduled the resource critical path.

At first, in the Activity gantt it looks like a mess. When you look at the Resource Gantt it makes more sense, so my first question in this is:

How did Spider decide which activities are first and wich are after?

The other thing I did was creating reference books for optimistic, most probable and pessimistic productivities. But when I tried to run the Risk Analysis, it gave me an error:

http://screenr.com/pwx

I don’t know how to correct that error. Any help?

I got some stress as I said that I would be delivering the schedule tonight but ... well, my purpose with this project is precissely making errors, and it is taking me time to understand how SDPM is calculated and how I am supposed to create a SDPM schedule.

Best regards,

Carlos.

Mon, 2010-02-22 13:32

#16
Carlos,

You can practice with the sample job “constr”.

Select Phase Line (if in “const” sample job select Construction works Phase), Right click and select Risk Analysis, Select your parameter and click OK.

From previous Vladimir assistance:

Beware that in the example that is in Demo the profit was calculated to some predefined date. That is why the duration is the same in all three versions. But if you will study the same for the phase Construction the dates and durations will be different.

To find the distribution curve you shall create three versions with different names (example: constr_opt, constr_mp, constr_pes). It does not matter in what folders they are kept (an example: Pessimistic on Server, Most Probable on Project Manager Computer, and Optimistic - on Project Planner computer.

When you first time select Risk Analysis in Calculations menu you shall define where is optimistic, most probable and pessimistic versions and the names of corresponding files. You will also set initial requirements for the probabilities to meet schedule, cost and other targets.

After risk calculation you will see Critical schedule and the distribution will be created. You can select if Critical Schedule will be based on optimistic or most probable estimates. We recommend most probable. Another option - to create backward Critical schedule not from the calculated finish but from tart\get finish dates. This option shall be used after defining targets.

Now you may see the distributions for the whole project and its elements in the pop-up menu (right mouse click on the row number). Here you can set target dates and costs.

You can also try the following software if interested in doing your own curve fittng.

http://www.mathwave.com/

Best regards,

Rafael

You can practice with the sample job “constr”.

Select Phase Line (if in “const” sample job select Construction works Phase), Right click and select Risk Analysis, Select your parameter and click OK.

From previous Vladimir assistance:

Beware that in the example that is in Demo the profit was calculated to some predefined date. That is why the duration is the same in all three versions. But if you will study the same for the phase Construction the dates and durations will be different.

To find the distribution curve you shall create three versions with different names (example: constr_opt, constr_mp, constr_pes). It does not matter in what folders they are kept (an example: Pessimistic on Server, Most Probable on Project Manager Computer, and Optimistic - on Project Planner computer.

When you first time select Risk Analysis in Calculations menu you shall define where is optimistic, most probable and pessimistic versions and the names of corresponding files. You will also set initial requirements for the probabilities to meet schedule, cost and other targets.

After risk calculation you will see Critical schedule and the distribution will be created. You can select if Critical Schedule will be based on optimistic or most probable estimates. We recommend most probable. Another option - to create backward Critical schedule not from the calculated finish but from tart\get finish dates. This option shall be used after defining targets.

Now you may see the distributions for the whole project and its elements in the pop-up menu (right mouse click on the row number). Here you can set target dates and costs.

You can also try the following software if interested in doing your own curve fittng.

http://www.mathwave.com/

Best regards,

Rafael

Sun, 2010-02-21 21:52

#17
Of course it easier to let the software find it using the Cumulative Distribution similar to Augustin’s on post#8? It can be for time, for costs or any other parameter available.

Sun, 2010-02-21 21:42

#18
I had made some errors.

I printed probability instead of productivity in one place and wrote about X instead of (1 - X).

X is the probability to be on time set by the project Target and is equal to 0.8 or 80%. If X is raising then the probability to meet duration target is raising.

Best Regards,

Vladimir

I printed probability instead of productivity in one place and wrote about X instead of (1 - X).

X is the probability to be on time set by the project Target and is equal to 0.8 or 80%. If X is raising then the probability to meet duration target is raising.

Best Regards,

Vladimir

Sun, 2010-02-21 21:19

#19
Sun, 2010-02-21 21:12

#20
Rafael,

we did not choose triangular distribution. I just suggest triangles for manual computations. The error does not matter much because initial data are not quite correct and precision is more necessary that an accuracy.

Carlos,

Rafael explained what is triangular distribution.

Let’s discuss your project.

If you spend 2.2 aeconds per word then your productivity is approximately 0.385 words per one second (please remember that in Spider project productivity is defined by volume per one hour).

Let’s assume that your optimistic probability 0.33 words per second and pessimistic is 0.44 words per second (plus or minus 0.55 words per second).

So your project optimistic duration is approximately O= 14,998 seconds, most probable M= 17,140 seconds and pessimistic P= 19,997 seconds.

Notice that I supposed that your productivity is Most Probable plus or minus the same value, but Duration deviations are not the same because it is Volume of Work (6,599 words) divided by productivity!

Now we can draw the triangle with the maximum at most probable value.

If we decided to set target duration that can be met with 80% probability we need to find what target value T will divide the triangle square in the proportion 2/8.

I asked Rafael to publish the picture that illustrates calculations but result can be calculated by the next formula:

1 – X = [(P-T)*(P-T)]/[(P-M)*(P-O)]

where X=0.2 in our case.

(I don’t know how to use superscript in this forum)

Target Duration is T=18307 seconds.

Later you shall calculate not target value but X. If X will become less than 0.2 then your performance is better than expected and your chances to be on time are increasing.

You shall always look ahead, your P, M, and O values will be new, only T will not change. This is one of the main differences between SDPM and other methods of performance management that make forecasts basing on past performance only.

Best Regards,

Vladimir

we did not choose triangular distribution. I just suggest triangles for manual computations. The error does not matter much because initial data are not quite correct and precision is more necessary that an accuracy.

Carlos,

Rafael explained what is triangular distribution.

Let’s discuss your project.

If you spend 2.2 aeconds per word then your productivity is approximately 0.385 words per one second (please remember that in Spider project productivity is defined by volume per one hour).

Let’s assume that your optimistic probability 0.33 words per second and pessimistic is 0.44 words per second (plus or minus 0.55 words per second).

So your project optimistic duration is approximately O= 14,998 seconds, most probable M= 17,140 seconds and pessimistic P= 19,997 seconds.

Notice that I supposed that your productivity is Most Probable plus or minus the same value, but Duration deviations are not the same because it is Volume of Work (6,599 words) divided by productivity!

Now we can draw the triangle with the maximum at most probable value.

If we decided to set target duration that can be met with 80% probability we need to find what target value T will divide the triangle square in the proportion 2/8.

I asked Rafael to publish the picture that illustrates calculations but result can be calculated by the next formula:

1 – X = [(P-T)*(P-T)]/[(P-M)*(P-O)]

where X=0.2 in our case.

(I don’t know how to use superscript in this forum)

Target Duration is T=18307 seconds.

Later you shall calculate not target value but X. If X will become less than 0.2 then your performance is better than expected and your chances to be on time are increasing.

You shall always look ahead, your P, M, and O values will be new, only T will not change. This is one of the main differences between SDPM and other methods of performance management that make forecasts basing on past performance only.

Best Regards,

Vladimir

Sun, 2010-02-21 19:59

#21
Carlos,

The x axis should represent the estimated durations, the y axis the probability for the estimated durations, the probability values can be determined from already available formulas based on the fact that the area under the curve equals one.

X.............Y.....................................

4.00 0.00

5.08 2/ (5.50-4.00) =1.33

5.50 0.00

http://www.riskamp.com/library/pertdistribution.php

There must be a reason why Spider Team chosed triangular over 3-point estimate. Vladimir is the one who can tell us. Maybe it is the best fit for the 3 models as the mode does not have to coincide with the median.

Best regards,

Rafael

The x axis should represent the estimated durations, the y axis the probability for the estimated durations, the probability values can be determined from already available formulas based on the fact that the area under the curve equals one.

X.............Y.....................................

4.00 0.00

5.08 2/ (5.50-4.00) =1.33

5.50 0.00

http://www.riskamp.com/library/pertdistribution.php

There must be a reason why Spider Team chosed triangular over 3-point estimate. Vladimir is the one who can tell us. Maybe it is the best fit for the 3 models as the mode does not have to coincide with the median.

Best regards,

Rafael

Sun, 2010-02-21 19:33

#22
Yes, but by that definition, what should be put in the X axis and what in the Y axis?

Hey Rafael, do you want to be the quality and style auditor for my translation Project?

There will be not money but ... maybe some learning.

Best regards,

Carlos.

Hey Rafael, do you want to be the quality and style auditor for my translation Project?

There will be not money but ... maybe some learning.

Best regards,

Carlos.

Sun, 2010-02-21 19:26

#23
Carlos,

From:

http://en.wikipedia.org/wiki/Triangular_distribution

“In probability theory and statistics, the triangular distribution is a continuous probability distribution with lower limit a, mode c and upper limit b.”

“The triangular distribution is typically used as a subjective description of a population for which there is only limited sample data, and especially in cases where the relationship between variables is known but data is scarce (possibly because of the high cost of collection). It is based on knowledge of the minimum and maximum and an "inspired guess" as to the modal value.”

From:

http://en.wikipedia.org/wiki/Three-point_estimation

“It differs from the double-triangular by its simple triangular shape and the mode does not have to coincide with the median.” The Three-point estimation assumes a symmetrical distribution.

Best regards,

Rafael

From:

http://en.wikipedia.org/wiki/Triangular_distribution

“In probability theory and statistics, the triangular distribution is a continuous probability distribution with lower limit a, mode c and upper limit b.”

“The triangular distribution is typically used as a subjective description of a population for which there is only limited sample data, and especially in cases where the relationship between variables is known but data is scarce (possibly because of the high cost of collection). It is based on knowledge of the minimum and maximum and an "inspired guess" as to the modal value.”

From:

http://en.wikipedia.org/wiki/Three-point_estimation

“It differs from the double-triangular by its simple triangular shape and the mode does not have to coincide with the median.” The Three-point estimation assumes a symmetrical distribution.

Best regards,

Rafael

Sun, 2010-02-21 17:45

#24
Rafael: Thanks for keeping this alive. I’m moving from Jalisco to Mexico DF so I have been a little bussy with all this moving thing.

keep reading this topic as I will be requesting some of your time as quality auditor.

Vladimir: Thanks for your time.

I can not fully understand something until I do it. So, I’m propossing a small, real project to be undertaken to understand SDPM.

I think a translation is a good project, it involves a skill, a fixed amount of work to be done and production rate easily measurable. Its only flaw is it is a linear project, but I think it is good as a first approach to SDPM.

So here’s the project: I will be translating the "First Project" file contained within Spider Project Demo help, I will work 1 hour daily translating from English to Espanol, and 1 hour into updating the schedule (I have good expectatives in Spider by the "Let’s challenge Spider" Topic, but I have not actually made a schedule in Spider). I will start this wednesday, I’m using from Monday morning to tuesday night to prepare a schedule.

In order to measure my own productivity at translating from English to Spanish, I translated the first paragraph, which is formed by 74 words. I finished it in 3:12.3, (192.3 Seconds), So I got a productivity of 2.598 seconds per word (Let’s kep it 2.6 Seconds). If the actual document contains 6,599 words, at 2.6 productivity I will be finishing it in 17157.4 seconds, that is 4.76 hours. This is where I will be needing a Quality auditor, i will be requesting some of Rafael’s time as he also speaks english-spanish (Please correct me if I’m wrong.)

I’m not understanding how a Triangular Distribution is used in PM, I think I should start by there before fully entering into SDPM. I drew this diagram trying to understand what you explained earlier:

"Draw the triangle distribution with the maximum at most probable value.

Now find the value that divides the triangle square at 4 to 1 proportion (4 to the left). Let’s set this duration value as the project target finish."

Please, may you correct me where I am wrong?

Please, share what you think about the project idea.

Best Regards.

Carlos.

keep reading this topic as I will be requesting some of your time as quality auditor.

Vladimir: Thanks for your time.

I can not fully understand something until I do it. So, I’m propossing a small, real project to be undertaken to understand SDPM.

I think a translation is a good project, it involves a skill, a fixed amount of work to be done and production rate easily measurable. Its only flaw is it is a linear project, but I think it is good as a first approach to SDPM.

So here’s the project: I will be translating the "First Project" file contained within Spider Project Demo help, I will work 1 hour daily translating from English to Espanol, and 1 hour into updating the schedule (I have good expectatives in Spider by the "Let’s challenge Spider" Topic, but I have not actually made a schedule in Spider). I will start this wednesday, I’m using from Monday morning to tuesday night to prepare a schedule.

In order to measure my own productivity at translating from English to Spanish, I translated the first paragraph, which is formed by 74 words. I finished it in 3:12.3, (192.3 Seconds), So I got a productivity of 2.598 seconds per word (Let’s kep it 2.6 Seconds). If the actual document contains 6,599 words, at 2.6 productivity I will be finishing it in 17157.4 seconds, that is 4.76 hours. This is where I will be needing a Quality auditor, i will be requesting some of Rafael’s time as he also speaks english-spanish (Please correct me if I’m wrong.)

I’m not understanding how a Triangular Distribution is used in PM, I think I should start by there before fully entering into SDPM. I drew this diagram trying to understand what you explained earlier:

"Draw the triangle distribution with the maximum at most probable value.

Now find the value that divides the triangle square at 4 to 1 proportion (4 to the left). Let’s set this duration value as the project target finish."

Please, may you correct me where I am wrong?

Please, share what you think about the project idea.

Best Regards.

Carlos.

Sat, 2010-02-20 11:05

#25
Vladimir,

Your suggestion will solve all issues with Owners if we keep the Most Probable as the official schedule for purpose of delays and as a true measure of what is “Most Probable”

Optimistic schedule is for workforce management. Most probable - for project management team.

This can provide me with what I need, this is no longer an issue. Now even if we do not track costs using the CPM but the Financial/Accounting software I will give it a try for purpose of keeping adequate buffer in our planning.

A hundred thousand thanks,

Rafael

Carlos,

Thanks for forcing the issue as I was kind of reluctant to get into it even when I knew all is automatically synchronized. While you can make your updates in any of the versions and the remaining two will be synchronized I will ask the field to update the Optimistic, the one they must target on the field.

Best regards,

Rafael

Your suggestion will solve all issues with Owners if we keep the Most Probable as the official schedule for purpose of delays and as a true measure of what is “Most Probable”

Optimistic schedule is for workforce management. Most probable - for project management team.

This can provide me with what I need, this is no longer an issue. Now even if we do not track costs using the CPM but the Financial/Accounting software I will give it a try for purpose of keeping adequate buffer in our planning.

A hundred thousand thanks,

Rafael

Carlos,

Thanks for forcing the issue as I was kind of reluctant to get into it even when I knew all is automatically synchronized. While you can make your updates in any of the versions and the remaining two will be synchronized I will ask the field to update the Optimistic, the one they must target on the field.

Best regards,

Rafael

Fri, 2010-02-19 12:29

#26
Augustin,

Even when some software can provide you with a way to simplify your estimate for each individual activity duration distribution the time spent on computations for thousands of activities and maybe hundred of thousand of activities in a Portfolio can easily make it impractical especially when Monte Carlo have to perform hundreds of schedule runs each with different activity durations determined from the Monte Carlo subroutine.

Also the fact that every software computes the schedule on its own way means that transferring data from a software non capable of Monte Carlo Simulation to another can invalidate the analysis. Again imagine transferring several portfolio jobs to a Monte Carlo simulation on every update. With Spider Project it is automatic and based on the same software model.

Critical Chain confusing? Of course, it tampers with classical computation of float representing a portion by some dummy activities, I have problems with that.

Best regards,

Rafael

Even when some software can provide you with a way to simplify your estimate for each individual activity duration distribution the time spent on computations for thousands of activities and maybe hundred of thousand of activities in a Portfolio can easily make it impractical especially when Monte Carlo have to perform hundreds of schedule runs each with different activity durations determined from the Monte Carlo subroutine.

Also the fact that every software computes the schedule on its own way means that transferring data from a software non capable of Monte Carlo Simulation to another can invalidate the analysis. Again imagine transferring several portfolio jobs to a Monte Carlo simulation on every update. With Spider Project it is automatic and based on the same software model.

Critical Chain confusing? Of course, it tampers with classical computation of float representing a portion by some dummy activities, I have problems with that.

Best regards,

Rafael

Fri, 2010-02-19 12:27

#27
Rafael,

our experience shows that most probable project duration has low probability to be achieved, usually between 25% and 38%. Monte Carlo simulations show real picture.

So we suggest to set project targets with 70%-80% probability to meet, and these targets do not belong to any schedule (except critical).

This is the problem with application of traditional performance management techiques - there is no baseline schedule that finishes on the target dates!!!

Of course it will not be accepted by your clients that do not support creating and using contingency reserves, but this is the life.

So we created additional option - to save the critical schedule that is close to what can be called schedule baseline.

But again - some of our contractors use this approach for risk management as a tool for contract management.

Pessimistic schedule is the contract schedule. Optimistic schedule is for workforce management. Most probable - for project management team.

Since all these schedules are synchronised it helps to manage these schedules in parallel and makes the contract management much more reliable. Parallel targets for different project participants are justified.

Critical Chain is a set of reasonable rules for manual management of simple projects. If you will look at these rules keeping this in mind you will not ask too much and will agree with the statements that are confusing.

Best Regards,

Vladimir

our experience shows that most probable project duration has low probability to be achieved, usually between 25% and 38%. Monte Carlo simulations show real picture.

So we suggest to set project targets with 70%-80% probability to meet, and these targets do not belong to any schedule (except critical).

This is the problem with application of traditional performance management techiques - there is no baseline schedule that finishes on the target dates!!!

Of course it will not be accepted by your clients that do not support creating and using contingency reserves, but this is the life.

So we created additional option - to save the critical schedule that is close to what can be called schedule baseline.

But again - some of our contractors use this approach for risk management as a tool for contract management.

Pessimistic schedule is the contract schedule. Optimistic schedule is for workforce management. Most probable - for project management team.

Since all these schedules are synchronised it helps to manage these schedules in parallel and makes the contract management much more reliable. Parallel targets for different project participants are justified.

Critical Chain is a set of reasonable rules for manual management of simple projects. If you will look at these rules keeping this in mind you will not ask too much and will agree with the statements that are confusing.

Best Regards,

Vladimir

Fri, 2010-02-19 12:01

#28
Hi Rafael,

Thank you for showing the cumulative distribution.

One huge difference between Three Scenario Approach and Monte Carlo is that for Monte Carlo application you need lots of historical data, lots of mathematics and justification for each choice you’ve made in the analysis, not mentioning the time spent on computations. It is a real research work.

For the Three Scenario Approach you need to identify the risks events and uncertainties, to prioritize them and to develop the risk response plan. Based on these you can easily include the risks events through their effect at the activity level into the pessimistic scenario and exclude most of them into the optimistic scenario.

Now you can justify better the risk environment the project will be executed. And the client will understand it easily too.

Best regards,

Augustin

Thank you for showing the cumulative distribution.

One huge difference between Three Scenario Approach and Monte Carlo is that for Monte Carlo application you need lots of historical data, lots of mathematics and justification for each choice you’ve made in the analysis, not mentioning the time spent on computations. It is a real research work.

For the Three Scenario Approach you need to identify the risks events and uncertainties, to prioritize them and to develop the risk response plan. Based on these you can easily include the risks events through their effect at the activity level into the pessimistic scenario and exclude most of them into the optimistic scenario.

Now you can justify better the risk environment the project will be executed. And the client will understand it easily too.

Best regards,

Augustin

Fri, 2010-02-19 11:16

#29
Vladimir,

Thanks, at first glance seems I got it, I will take a deeper look at it latter. Maybe Critical Chain application of buffers where you assign them arbitrarily got me confused.

Now I understand buffer is a computed value, a comparison value between two schedule versions, and was missing that the optimistic schedule should be used for day by day management and not the most probable. You got to target/plan optimistic schedule in order to get on or before most probable durations schedule.

This is in agreement with Monte Carlo Simulation that usually shifts the distribution curve to the right of the “most probable durations” schedule.

Best regards,

Rafael

Thanks, at first glance seems I got it, I will take a deeper look at it latter. Maybe Critical Chain application of buffers where you assign them arbitrarily got me confused.

Now I understand buffer is a computed value, a comparison value between two schedule versions, and was missing that the optimistic schedule should be used for day by day management and not the most probable. You got to target/plan optimistic schedule in order to get on or before most probable durations schedule.

This is in agreement with Monte Carlo Simulation that usually shifts the distribution curve to the right of the “most probable durations” schedule.

Best regards,

Rafael

Fri, 2010-02-19 09:07

#30
Rafael,

you can define a set of target dates and costs (for the project as whole and for selected phases) and calculate initial probabilities to meet set targets.

Project (phase) time buffer is the difference between target and scheduled finish dates. It does not belong to any chain at all!!!

We recommend to use optimistic schedule for day by day management. Spider Project calculates Critical Schedule - this is the schedule (usually with most probable activity durations) calculated backward from the target dates. And Spider shows time reserves for all activities as the difference between their finishes in Critical and Current schedules. If you reached Critical schedule then you are in serious trouble.

Of course time buffers are consumed and there is a need to estimate if this consumption is larger or lower than was planned. Success probabilities that are calculated for all project targets measure buffer consumption. If success probability dropped then your buffer was consumed faster than expected or new risks that were included in pessimistic scenario made your buffer less reliable. Success probabilities show project (phase) health.

But even more information you can get from success probability trends. You will be able to discover the problems early. If current success probability is relatively high but the trend is negative then something goes wrong and you shall do something to improve the performance. If the trend is positive you can conclude that the performance is improving and even low success probability don’t require immediate actions.

The same with costs, material consumptions and other targeted parameters.

So project (phase) buffers are project contingency reserves. Time buffers may be seen at the Gantt Chart as the difference between Critical and current scheduled dates, and in columns of the Gantt Chart table. They do not prevent to calculate real floats and real costs, they are not represented by some dummy activities.

Best Regards,

Vladimir

you can define a set of target dates and costs (for the project as whole and for selected phases) and calculate initial probabilities to meet set targets.

Project (phase) time buffer is the difference between target and scheduled finish dates. It does not belong to any chain at all!!!

We recommend to use optimistic schedule for day by day management. Spider Project calculates Critical Schedule - this is the schedule (usually with most probable activity durations) calculated backward from the target dates. And Spider shows time reserves for all activities as the difference between their finishes in Critical and Current schedules. If you reached Critical schedule then you are in serious trouble.

Of course time buffers are consumed and there is a need to estimate if this consumption is larger or lower than was planned. Success probabilities that are calculated for all project targets measure buffer consumption. If success probability dropped then your buffer was consumed faster than expected or new risks that were included in pessimistic scenario made your buffer less reliable. Success probabilities show project (phase) health.

But even more information you can get from success probability trends. You will be able to discover the problems early. If current success probability is relatively high but the trend is negative then something goes wrong and you shall do something to improve the performance. If the trend is positive you can conclude that the performance is improving and even low success probability don’t require immediate actions.

The same with costs, material consumptions and other targeted parameters.

So project (phase) buffers are project contingency reserves. Time buffers may be seen at the Gantt Chart as the difference between Critical and current scheduled dates, and in columns of the Gantt Chart table. They do not prevent to calculate real floats and real costs, they are not represented by some dummy activities.

Best Regards,

Vladimir

Fri, 2010-02-19 06:08

#31
Amazing, I would be concerned about contractual milestones. Seems like easy to do, quite good.

What about buffer protection for achieving the individual milestones, the "Finish Buffer" values of activities on the milestones individual path?

Can you give me a conceptual definition of "Finish Buffer" other than the matematical? Is it something like a "reserve float" to achieve target date within a given probability? Is it related to the overall job longest path?

Best regards,

Rafael

What about buffer protection for achieving the individual milestones, the "Finish Buffer" values of activities on the milestones individual path?

Can you give me a conceptual definition of "Finish Buffer" other than the matematical? Is it something like a "reserve float" to achieve target date within a given probability? Is it related to the overall job longest path?

Best regards,

Rafael

Fri, 2010-02-19 02:16

#32
We can get the distribution for the finish or cost of any project phase.

Usually it is not necessary to learn the distribution for individual activities. In any case you can always create a phase where this activity belongs. Don’t forget that the number of parallel WBS is unlimited.

Best Regards,

Vladimir

Usually it is not necessary to learn the distribution for individual activities. In any case you can always create a phase where this activity belongs. Don’t forget that the number of parallel WBS is unlimited.

Best Regards,

Vladimir

Thu, 2010-02-18 19:17

#33
Vladimir,

If the metrics make sense for the hole project longest path, then what about individual milestones or activities. Can we get the cumulative distribution for the finish of individual milestones and individual activities? And about finish buffer relative to individual milestone or activities. I believe the data is already there.

Best regards,

Rafael

If the metrics make sense for the hole project longest path, then what about individual milestones or activities. Can we get the cumulative distribution for the finish of individual milestones and individual activities? And about finish buffer relative to individual milestone or activities. I believe the data is already there.

Best regards,

Rafael

Thu, 2010-02-18 15:29

#34
Thu, 2010-02-18 13:58

#35
I tried to include a link to a graphic. I miss something.

Augustin

Augustin

Thu, 2010-02-18 13:57

#36
Hi Rafael,

I agree you with you. Not all the clients understand the probabilities, but they are concerned on risks and on their impact.

I usually show them the probability curve after I explain the impact of different risk events and uncertainties. This is easier for them.

http://i833.photobucket.com/albums/zz253/augustin_purnus/Graphic1.jpg

After I play with the target date or with the current probability, they foresee the benefits of such approach. Of course, not all of the clients are opened to such approach. It depends on their “culture”.

Best regards,

Augustin

I agree you with you. Not all the clients understand the probabilities, but they are concerned on risks and on their impact.

I usually show them the probability curve after I explain the impact of different risk events and uncertainties. This is easier for them.

http://i833.photobucket.com/albums/zz253/augustin_purnus/Graphic1.jpg

After I play with the target date or with the current probability, they foresee the benefits of such approach. Of course, not all of the clients are opened to such approach. It depends on their “culture”.

Best regards,

Augustin

Tue, 2010-02-16 21:10

#37
Carlos,

The Critical Path activities between the scenarios does not have to be the same, a mathematical error of original PERT, was solved with Monte Carlo Simulation but this Spider Project approach deals with the issue very efficiently.

You create the 3 scenarios by duplicating you base schedule and making the adjustments then at every update the software synchronizes all scenarios at the click of the mouse.

My issue is with my clients, if you start talking to them about probabilities they will get scared. Some advanced features are not for everyone. In order for the scenarios to be real you need the input of your clients. If your clients have no issue with probabilities and keep track of actual costs and resource usage then you better get into it.

Best regards,

Rafael

The Critical Path activities between the scenarios does not have to be the same, a mathematical error of original PERT, was solved with Monte Carlo Simulation but this Spider Project approach deals with the issue very efficiently.

You create the 3 scenarios by duplicating you base schedule and making the adjustments then at every update the software synchronizes all scenarios at the click of the mouse.

My issue is with my clients, if you start talking to them about probabilities they will get scared. Some advanced features are not for everyone. In order for the scenarios to be real you need the input of your clients. If your clients have no issue with probabilities and keep track of actual costs and resource usage then you better get into it.

Best regards,

Rafael

Tue, 2010-02-16 16:08

#38
Hi Carlos,

yes, you can apply SDPM manually but it is time consuming because you shall create three scenarios of your project.

Create optimistic, most probable and pessimistic scenarios (schedules) and calculate optimistic, most probable and pessimistic durations.

Draw the triangle distribution with the maximum at most probable value.

Now find the value that divides the triangle square at 4 to 1 proportion (4 to the left). Let’s set this duration value as the project target finish.

Initial probability to finish earlier is 80% (very approximately but it does not matter much).

Optimistic schedule shall be used for setting tasks to project stuff. Your buffer is the difference between the target and optimistic durations.

Now simulate the performance. Assume that some works were done and enter their actual information in all three schedules. Calculate new durations. Create the same triangle and calculate the square at the left of the target duration. Is it now larger or smaller than 80%?

If smaller then your performance was not good and your buffer consumption is larger than expected.

Of course this is the simplification but may help to understand the approach that shall be applied not only to durations but also to costs and material consumption.

The idea is clear and it works perfectly but for the real projects there is a need in the software tools.

Best Regards,

Vladimir

yes, you can apply SDPM manually but it is time consuming because you shall create three scenarios of your project.

Create optimistic, most probable and pessimistic scenarios (schedules) and calculate optimistic, most probable and pessimistic durations.

Draw the triangle distribution with the maximum at most probable value.

Now find the value that divides the triangle square at 4 to 1 proportion (4 to the left). Let’s set this duration value as the project target finish.

Initial probability to finish earlier is 80% (very approximately but it does not matter much).

Optimistic schedule shall be used for setting tasks to project stuff. Your buffer is the difference between the target and optimistic durations.

Now simulate the performance. Assume that some works were done and enter their actual information in all three schedules. Calculate new durations. Create the same triangle and calculate the square at the left of the target duration. Is it now larger or smaller than 80%?

If smaller then your performance was not good and your buffer consumption is larger than expected.

Of course this is the simplification but may help to understand the approach that shall be applied not only to durations but also to costs and material consumption.

The idea is clear and it works perfectly but for the real projects there is a need in the software tools.

Best Regards,

Vladimir

Tue, 2010-02-16 13:47

#39
Hi Vladimir.

I am undersanding, is it necessary the use of a software to sucesfully make SDPM?

The sense of my question was more on the side of "Learn SDPM without using software", like when I learnt CPM with a papersheet and a pencil. The Software just automates the process, right? In any case, if I can’t avoid using software, I will be using Spider Project Demo. I see no point in using Spider for CPM if it was built to make SDPM. It is not that I don’t want to use software when managing a project, is that I want to understand why the software works the way it does.

Thanks for the link. I’m reading topics there, I will be posting more questions at this topic as they arise.

Best Regards,

Carlos

I am undersanding, is it necessary the use of a software to sucesfully make SDPM?

The sense of my question was more on the side of "Learn SDPM without using software", like when I learnt CPM with a papersheet and a pencil. The Software just automates the process, right? In any case, if I can’t avoid using software, I will be using Spider Project Demo. I see no point in using Spider for CPM if it was built to make SDPM. It is not that I don’t want to use software when managing a project, is that I want to understand why the software works the way it does.

Thanks for the link. I’m reading topics there, I will be posting more questions at this topic as they arise.

Best Regards,

Carlos

Sun, 2010-02-14 04:08

#40
Hi Carlos,

SDPM is the methodology.

Spider Project supports this methodology and at my presentations I used Spider screenshots for illustrations.

With more efforts it can be used with other software.

Please inform me what software do you use and we will discuss how to apply SDPM with your software.

Visit site www.sdpmworld.com that is recently created. This site will be developed and materials and discussions on SDPM will be added to its current content.

SDPM has common features with the Critical Chain but was developed independently (and earlier). There are many differences too. I consider CC as a set of reasonable instructions for those who manage small projects manually (without proper software tools).

Spider Project supports CPM as well. SDPM is the next project management maturity level.

Best Regards,

Vladimir

SDPM is the methodology.

Spider Project supports this methodology and at my presentations I used Spider screenshots for illustrations.

With more efforts it can be used with other software.

Please inform me what software do you use and we will discuss how to apply SDPM with your software.

Visit site www.sdpmworld.com that is recently created. This site will be developed and materials and discussions on SDPM will be added to its current content.

SDPM has common features with the Critical Chain but was developed independently (and earlier). There are many differences too. I consider CC as a set of reasonable instructions for those who manage small projects manually (without proper software tools).

Spider Project supports CPM as well. SDPM is the next project management maturity level.

Best Regards,

Vladimir

## Replies