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.

Risk analysis

54 replies [Last post]
Pramesh K W
User offline. Last seen 15 years 20 weeks ago. Offline
Joined: 19 Jun 2003
Posts: 10
Dear Planners ,

Can anybody throw some light on easier methods of risk analysis thru excel ?

Has anybody worked on actual live projects and done schedule risk analysis ?

Regards

Replies

Rahul Mulik
User offline. Last seen 10 years 4 weeks ago. Offline
Joined: 24 Dec 2005
Posts: 15
Groups: None
Dear all,
Currently I am working in Buildings Works, the list for the RED RISK or RISK PRONE or ALERT ZONE events/activities in Building Construction Projects, according to me would be,

1. Approvals & Clearances by Concerned Govt Authorities,
2. Sign Off of Concept Design for Internal Finishes,
3. Sign Off of Architectural Drawings (Design-Built),
4. Subcontractor Identification & Appointment,
5. Material & Drawing Submittals Approvals by Project Management Consultant,
6. MEP Items Procurement (When requirements are customised)

Most of the items in the list would be form Design Development & Procurement Package of the Project Life Cycle.

I would like the amendments and comments on the list.

Cheers & Regards,
Rahul
Norzul Ibrahim
User offline. Last seen 11 years 44 weeks ago. Offline
Joined: 17 Dec 2005
Posts: 165
Dear Jerry,

Welcome to the forum? Where are yu know? Long time no see....he he he....

Thanks

norzul
Jerry Alivio
User offline. Last seen 11 years 13 weeks ago. Offline
Joined: 6 Aug 2005
Posts: 56
Dear All,

I certainly agree with Mr. Salamat, Risk in schedule is just like an Aplha & Omega. In oil & gas plants and refineries especially plant shutdown & Turnaround projects i always meet this interfacing of activities most are from piping works, asbuilt.

Jerry Alivio

Raja Izat Raja Ib...
User offline. Last seen 8 years 12 weeks ago. Offline
Joined: 2 Jun 2005
Posts: 299
Hi Pedrito,
agreed with your post and additional info to be include like
1. Local Authorities needs before execute project.
2. Access, Utilities, Environment.
3. Contractors, skill workers ,Labour available at the countries.
4. Inspection relate with Local Authorities.
This is some common points we can highlight, actually there is so many points but its depends on the project.
Pedrito Salamat
User offline. Last seen 8 years 21 weeks ago. Offline
Joined: 12 Jul 2005
Posts: 17
Groups: None
Dear All,

Risk is always part of the planning and you will always encountered in Oil & Gas or Petrochem Construction due to the following reasons particularly for FEED or EPC Contract.
1) Starting from the concept design where Plant has to meet the specification requirement by the Client.
2) Developing definitive design based on the approved concept with the concurrence or approval from the Consultant or the expert
3) Approval of the materials and Procurement where they requires time for the approval of documentation, manufacturing period and delivery from the origin.
4) Other factors or site condition that might affects the design where deviation from specs should require.
5) Interfacing of the above items to the construction that will define the Critical Path.
6) Site working condition if there is the existing Plant and requires Interfacing of the system.
7) Plant Pre-commissioning work where system might later discovered some under design that will replace equipment.
8) And finally, the Plant Start-up whether the finish product really meets the specification demanded by the Client.

I have to note that these sequence of events will be typically always at risk , if you have no any proper planning.
Charleston-Joseph...
User offline. Last seen 7 years 44 weeks ago. Offline
Joined: 10 Jul 2005
Posts: 1347
Groups: None
Hi,

It is preferable to use crystal balls.

FEEDS to EPC, there is nothing there, all information are insufficient.

RISK IS ALWAYS THERE, INHERENT IN THIS TYPE OF WORLD.

But then petrochmem, oil and gas are unique breed.

Until now i wonder why i learn - dia inch.

We don’t have this in building works, heavy civils or infra works.

Cheers, (the experience in working at petrochem, oil and gas was worth the time, it is just not my career direction)
Ruel Llegado
User offline. Last seen 13 years 4 weeks ago. Offline
Joined: 28 Sep 2005
Posts: 8
In all projects, there will always be a "Risk". Mr.Salamat is correct about good planning. It is very essential that from the very beginning of the project, we must establish a good, realistic with resource loading schedule.

But on the FEED+EPC+Performance Testing Project, we can only apply going outside to downstream of construction (incl. pre-comm/comm/testing)to know and physically quanitify progress. However, FEED + EP, we should atleast look for all the documents and deliverables to the Document Data Base and Delivery Data Base to update progress.

Furthermore, Risk analysis can only be done (in FEED, EP)if you have the ontime information, correct sequence of activities, deliverables equivalent to Act. ID, etc...

But the best tools is "Experience" on how to detect risk and its countermeasures.

I hope this will help.

Charleston-Joseph...
User offline. Last seen 7 years 44 weeks ago. Offline
Joined: 10 Jul 2005
Posts: 1347
Groups: None
Hi to all,

I did like the discussion on risk analysis. It looks complicated due to the tools used in risk analysis.

In managing multiple projects, there is alway an inherent risk due to factors that influence the progress of the work. Some cannot be identified at the planning stage while some maybe specified already in the contract.

What is the best approach and user friendly remains a question?

Cheers,

Charlie
Pedrito Salamat
User offline. Last seen 8 years 21 weeks ago. Offline
Joined: 12 Jul 2005
Posts: 17
Groups: None
Risk is always there from start to the end. The only way to be out of the risk is the good planning.

Ensure that you have visited and evaluated the work site condition, identify logical work sequence and perform the lowest detail of planning with proper interfacing of work.

While to deal really with risk, dig up all the options and decide wisely to ensure that you can contain these rsik.
Daya Sugunasingha
User offline. Last seen 11 years 23 weeks ago. Offline
Joined: 27 Feb 2002
Posts: 99
Groups: None
This is what is so good about a forum like this!
There is always something to learn about the facts and views as seen from another industry. Thank you.
Frank Borcherdt
User offline. Last seen 2 years 35 weeks ago. Offline
Joined: 3 May 2001
Posts: 88
Groups: None
Philip,

No argument...the only absolutes are birth, death and taxes.

I was responding to the posts indicating that somehow risk management is about getting out and visiting the site.

As you know, being Petro-Chem, no amount of site visits prior to the inspection activities during the shutdown is going to narrow the risk beyond what can be established from the historical/statistical records on those pieces of equipment.

Petro-Chem usually requires developing a risk model that at least quantifies the worst case and allows you to load the schedule with activities that are desirable but could be discarded and thus provides the necessary level of resource to accommodate the worst case scenario.

I rarely see this approach adopted in other industries, but then again most are not continuous process in nature. I also have not seen too many shutdowns where attempts are made to perform risk analysis once the shutdown has started.
Philip Rawlings
User offline. Last seen 3 years 30 weeks ago. Offline
Joined: 16 Nov 2004
Posts: 41
Groups: None
Frank
Dont you belive it!
Whilst it is true that the greatest scope for mitigating risk is in the early stages where alternatives may be considered, the risks dont go away till the end. Obviously, good planning (and good execution) will reduce risk impacts there may still be risks that are not so easily reduced. Also, the better scoped the project is the better risks can be avoided - but, as ever, there are no absolutes.
Frank Borcherdt
User offline. Last seen 2 years 35 weeks ago. Offline
Joined: 3 May 2001
Posts: 88
Groups: None
Whilst I agree that there is no substitute for site visits once the job is underway, it does yield benefits during the initial feasibility/tender/establishment stage.

Once the detailed planning is complete in a well scoped project much of the risk is mitigated.
Bill Guthrie
User offline. Last seen 3 years 14 weeks ago. Offline
Joined: 14 Mar 2005
Posts: 262
Philip hit it on the head.

Get out in the field and look at the job.
Best Way.

And really gentlemen, Risk Analysis is basically you get what you put into it. You can basically factor in anything you want.

Have seen Risk alalysis completely change, basically as if you did not like the answer, you changed the questions. Anyone can force a risk analysis to say what they WANT it to say.
Some contractors ask what we want and they talor make a risk analysis to say want we want to hear. welcome to the real world

therefore, back to square one, is it really a risk involved, go out and look.


Regards Bill
clarence choi
User offline. Last seen 8 years 6 days ago. Offline
Joined: 28 Feb 2004
Posts: 6
Groups: None
Alex,

Can’t agree more.
Bear in mind the first step of any risk analysis is to identify risk.
If you have no idea what the risks are, how can you interpret the result?
Alex Wong
User offline. Last seen 6 years 21 weeks ago. Offline
Joined: 12 Feb 2003
Posts: 874
Groups: TILOS
Philp

I agree with you the part that we"Planner" should go out and see more rather than sitting in the office and trying to calculate upto the 2 decimal place of the risk.

Remember, computer can count number, we can make decision base on feel. The computer tool in front of us will clarify some of the feeling to fact. But how accurate you want it. It is up to individual to balance between them. Would you like to make the wrong decision on the right time or make the right decision at the wrong time. Remember schedule is not a precision tool.

Hope someone can share the same view.

Cheers

Alex
Daya Sugunasingha
User offline. Last seen 11 years 23 weeks ago. Offline
Joined: 27 Feb 2002
Posts: 99
Groups: None
I quite agree
Philip Jonker
User offline. Last seen 10 years 37 weeks ago. Offline
Joined: 7 Nov 2004
Posts: 852
Groups: None
I think you guys are missing the point, get up of your backsides, leave the computors for a bit, and see what is going on. Stop trying to figure out iterations, and the rest, go out and see what the problem really is, and try and help to solve it, that is planning.
Philip Rawlings
User offline. Last seen 3 years 30 weeks ago. Offline
Joined: 16 Nov 2004
Posts: 41
Groups: None
Robert
If you are recommending this, then a reference might be useful (for me and others).
Philip Rawlings
User offline. Last seen 3 years 30 weeks ago. Offline
Joined: 16 Nov 2004
Posts: 41
Groups: None
Hello everyone
I think the point is being missed here a little regarding the (absolute) number of iterations required. If you wnat no variation between different runs (on different software, computers, seeds) then I suggest an infinite number of iterations.

In the real world there is no real answer. The practical answer depends on the particular model and mostly on the nature of the critical paths. However many activities and risk values you have, if there is a dominant critical path, then the number of activities modelled is effectively the number of activities on that path (or paths).

Although there is no theoretical, general answer I would suggest that you just run the risk analysis again with a different seed (starting point for selection of random nymbers). If the answer is unacceptably different (at mean or P50 or whatever you are interested in) increase the iterations. What is an acceptable difference? That which would change any decisions/strategies you make based on the risk analysis.
Stephen Devaux
User offline. Last seen 2 hours 35 min ago. Offline
Joined: 23 Mar 2005
Posts: 619
Vladimir wrote:

"The question is open - how many iterations are necessary for Monte Carlo simulation of 10000 activities resource loaded project schedule? . . Does anybody know the answer?"

What is your guess, Vladimir? I’m really interested.

"I mean receiving the probability estimations with not more than 1% difference if to do Monte Carlo simulation once more"

Can I suggest that the variables involved in the input to a Monte Carlo-type system, and the work performance, cause a MUCH greater noise than algorithmic variances?

I’m also curious if we’re talking specific distribution shapes for each activity estimate, or simply a "standard" Beta or triangular distribution. How much of a difference does that make? Obviously, different distibution shapes will result in very different outputs. You think that the 85% estimate is too long? Try changing from triangular to Beta distribution!
Ravi Potwar
User offline. Last seen 4 weeks 3 days ago. Offline
Joined: 7 Mar 2005
Posts: 29
Groups: None
Hi ,
Can some body ? who have done it ,please send some Sample project /Excel Sheet & guide to understand the method how to calculate & Measure the Risk analysis. I am New to this subject & like to learn more ,
Thanks
Of course it is necessary to do everything that is needed for risk avoidance or mitigation but still you will not be able to avoid them all. Besides your estimations of activity durations, costs, resource requirements are not perfect, there is always some uncertainty. Simulating risks you can set reliable targets, determine necessary contingency reserves and control success probability trends.
Risk simulation is necessary for reliable estimates of future project results.
The question is open - how many iterations are necessary for Monte Carlo simulation of 10000 activities resource loaded project schedule?
Assumptions - no risk events and conditional branches, no calendar uncertainties, just uncertainties in estimating every activity duration and cost. Additional requirement - 1% accuracy of process convergence. I mean receiving the probability estimations with not more than 1% difference if to do Monte Carlo simulation once more.
Does anybody know the answer?

To Serge Marashlian:
There is always some uncertainty in estimating any activity duration. It does not mean that project planning is useless. Proper project planning takes these uncertainties into consideration to determine which contingency reserves are necessary. The schedule targets that were determined without comsidering uncertainties and risks have low probability to be met.
Philip Jonker
User offline. Last seen 10 years 37 weeks ago. Offline
Joined: 7 Nov 2004
Posts: 852
Groups: None
Hi Serge/Tomasz,

You guys are practically hittting the nail on the head.

Our job as planners is to be pro-actively be managing the risks, and ensure they do not occur. How is this done? Not by being a smart arse computor operators, but by being pro-active and using our engineering experience to help the project. We can sit on this web site all our lives, and talk about claims/computor problems and other such things, and still get nowhere. The point is understand the risks and manage them.

I gad an instance last month where I showed 16% progress for the month, and I got a report back, that this was impossible, the contractor actually achieved 43%. The point is to be practical. If you have a three month contract you have to achieve 40% per month, despite it all
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Serge :-)

Your assumption is perfect!!! This assumption is really philospophy of Project Management, specially risk Management. In practice all factors and processess are risky but manager must to reduce this risk - radically (!). But total reduction is not possible in 100%, of course. (!!!). More suitable is control of risk - not simulating it - but before selection of "risk management method" we must to evaluate levels of risk. If we run random tests without any rational types of distributions our selection/evaluation is bad and our risk analysis is not valuable (it is only play/fun). I suppose that this discussion is related with serious problem of valuability/rationality of risk analysis because discussioners want to be serious managers/planners meybe analysers or simply fanatics of good/faithful planning.

Hello Everyone,
Tomek from Poland
Serge Marashlian
User offline. Last seen 5 years 10 weeks ago. Offline
Joined: 18 Apr 2005
Posts: 6
Groups: None
Hello Everyone

While reading your postings, i asked myself why do we need to determine the number of iterations to be allocated for a certain schedule. I am not a risk analyst but i can if possible tell you that we (planners) use the planning softwares in order to predict actions in the future and for sure in construction industry it can not be possible that the 10000 activities need to have a risk analysis. Since if that is the case it is better for all of us to pack and leave (the budget would not allow to have so many un-determinable activities)

Now if we need to transfer the risk analysis to a construction schedule, i think identifiying the risky items could be (if you agree) long lead material delivery, new method in construction or availability in a labor skill. My question is how do you determine a probability distribution for a certain activity (bell-shape, constant, skeweed etc).

Your feedback is highly appreciated
Serge
Yes, Frank.
That is why I asked a simple question. You have 10000 activities schedule, no prbabilistic branches, but these activities have uncertain durations and costs (let’s suppose that you have optimistic, most probable and pessimistic estimates and suppose beta-distribution on each activity). Activitites are linked, schedule is resource loaded and resources are limited. Let’s forget about calendars and risk events to make the task simple.
So the question - what is the number of necessary iterations. It is pure mathematical question and I hope that those who use Monte Carlo simulation know the answer.
Vladimir
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Vladimir,

Yes, from first point of view 5.000 iterations for 4.000 tasks is not good but in this shedule only part of tasks was randomised because other tasks was not risky. In this schedule I have meybe 500 usual tasks and rest tasks are "aiding" tasks, so relation of iterations/tasks was not bad :-) I have used pre-tests and 10.000 runs was not better. If we know process (!) we can conclude from small experiments.

In the aspect of number of iterations (replications) I suppose that valuable method is something like LasVegas simulation, i.e. running of testing runs with different numbers of iterations. If this base test researched parameter is for f.e. duration of project - and if we have smoother probability function of this duration our number of runs is "good".

Other method is using of convergension option (option in software system) i.e. simulating of schedule until test rule is fulfiled. This test is f.e. maximal variance of one tested parameter.

Of course this methods (manners) are not scientifically perfect because in model we can research not only one parameter (f.e. duration of project) and we have big number of complex variables, not only number of tasks. In my oppinion this discussion may be good if we want to calculate number of runs in the function of ONE parameter, f.e. number of tasks. In opposit situation any number is not right :-)) because we don,t know what we have in the model!

sorry for my not detailed question :-) I suppose that actual question is better.

all the best
dr Tomek :-)
Frank Borcherdt
User offline. Last seen 2 years 35 weeks ago. Offline
Joined: 3 May 2001
Posts: 88
Groups: None
The number of iterations to reach convergence is not a function of the number of activities but a function of the complexity of the problem eg number probabilistic and conditional branches, use of probabilistic calendars, etc, etc.
Dear Tomasz,
I don’t agree that Monte Carlo simulation for summary activity has sense - results will be certainly wrong if your resources are limited. Not only task existence but also task duration and cost shall be simulated.
I did not intend to estimate and to compare the calculation speed of any software tool. I believe that Pertmaster is faster than old Monte Carlo. It does not matter in this discussion. But you shall admit that 5000 iterations for 4000 activities project is just a fraction of the necessary number of iterations for reliable results. I am sure that if you will run this number of iterations on several computers then the results of Monte Carlo simulation will be different (assuming that all project activities have uncertain durations and costs).
A small remark about highest on the earth P3 power.
Recently I calculated project schedule for 2700000 activities project with limited resources (resource leveling) using Spider Project. So your information is not quite correct.
Thank you for your clarification but still there is a question - what is the necessary number of iterations for the project consisting of N activities for receiving reliable results (it does not depend on the software that is used). If somebody doesn’t know the answer for this question then he or she is not able to use Monte Carlo simulation properly

Regards,
Vladimir
Scott,
unfortunately you received information that shall be verifyed. It is not that easy, they did not tell you that analyzing large schedules you need to make millions of iterations to receive reliable results. It means that calculating reliable probability curve for the large project may take years. Usually people have not that much time and restrict the number of iterations by several hundreds. It means that calculated probability of meeting some target is not correct. Let’s suppose that an error is within 10% and it is enough for initial estimate. Now let’s suppose that later you decided to analyze project status and calculated the same probability once more. If the result shows that it became 5% lower you will not be able to make any conclusion. It may be caused by project performance problems or by unsufficient number of iterations.
Unfortunately you will not be able to estimate project performance.
It is interesting that people promoting Monte Carlo simulation never answered these questions - what number of iterations is necessary for receiving reliable probability curve for 10000 activities project (if each activity has uncertain duration and cost) and what time is necessary for proper Monte Carlo simulation of this project. Another question that may be asked - how to estimate the reliability of obtained results (error’ size) if you had made only 0.01% of the necessary number of iterations?
I hope that Tomasz will answer. Being Pertmaster and MC fanatic he shall know proper answers.
Mabe somebody else will join this discussion.
Regards,
Vladimir
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Vladimir :-)

Monte Carlo simulation for summary activities HAVE sense but not in the same form like on activities, of course. In Pertmaster or Project we can define dependencies on the summary tasks level (very good), not only on activity levels – Dear Vladimir (I do it). Of course, main purpose of summary tasks in “summarization” of subtasks, so defining of risk for summary tasks is not valuable, but we can do it (only probability of task existence but we can analyse risk related with summary task – this have sense). Meybe my e-mail was not clear.

If You write about time of simulation (this question was not contained in source e-mail :-) I know that simulation is speeder in Pertmaster than in systems P3+MCforP3. In aspect of 10.000 tasks project :-) I have no experience sorry. Duration of this simulation need probably less than ~ 30 min, but it will be related with power of computer!!! Meybe 2 years ago I have run simulation of 4.000 tasks project with highly complicated resource structure (file 5MB) on old notebook with Pentium 1 MMX 233MHz and simulation with 5.000 iterations was speeder than 30 min (after making free place on hard disk). I don’t see any problem with simulation of bigger projects, f.e. 100.000 tasks but in practice 100.000 of tasks it is highest level of P3 power (highest on the earth) and this level is related with all project portfolio in big firm and on this level of PM risk simulation is not used, I suppose.

In other case I have run (with students) the same file (5MB) on 8 machines with 5000 iterations (on every machine) but supply unit of server was "burn down" and our simulation was abandoned :-)). But this problem was related with hardware, not with software. New network server have new supply unit ;-) I suppose that simulation is not complicated for modern systems, but I not think that these systemns are ideal of course ;-)

Regards,
Vladimir

P.S. If other readers are interested in detail comparison of simulation speed of P3+MC and Pertmaster I would like to commend him good article of Philip Rawlings in PM Today. He said that Pertmaster is maybe 100x speeder than P3+MC. It is question related strictly with questions of Scott and Your famous person ;-)). Good Luck for all experts!
Scott Tate
User offline. Last seen 10 years 44 weeks ago. Offline
Joined: 4 Mar 2005
Posts: 28
Groups: None
A friend suggested I contact Pertmaster, which I did and learned what you stated, that the logic is necessary. Also, Pertmaster recognizes up to 128,000 activities and it is just as easy for the user to have Pertmaster analyze a 10,000 activity schedule as it is a 1,000 activity schedule. The only additional time is waiting for the computer to generate its analysis. I only learned of Pertmaster today and did not realize it had replaced the old Monte Carlo.
Tomasz,
you did not answer very interesting questions asked by Scott.
I will ask them again though in other way:
1) what is the necessary number of iteratiions for producing reliable probability distribution for the schedule consisting of 10000 activities?
2) how long does it take?

Scott,
Monte Carlo simulation for summary activities does not make sense because logical dependencies are defined on the activity level.

Regards,
Vladimir
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Scott and Dayanidhi,

I know that Primavera Systems don’t improve MonteCarlo system today and this system is not connected with new Primavera Enterprise (only with older P3). I know that today official risk analysis software for all products of Primavera is Pertmaster, not MonteCarlo for P3.

I am fanatic of Pertmaster (Primavera of course too) not only in newest form of Pertmaster with MonteCarlo engine. But If you are interested in MC method I would like to inform that MC is here great, powerfull improvement of schedule analysis because You can calculate easy all probabilities of important aspects of schedule (You know it of course). We have not any limitations of MC analysis, I suppose :-). It is not GERT of course, but You can model different interesting situation with good effects! Some problems we have with interpretation of some interrelations but it is subject for other, deep discussion (I will write about it in PEO).

I don’t have license of old MC for P3 but in Pertmaster we can model the risk for summary and hammock tasks (in P3 we have not summary tasks, of course) but risk is related specially with "child" tasks, not with summary tasks. In my oppinion comparisson of MC for P3 with MC in Pertmaster is not valuable today. In some aspects MC for P3 was perfect/best but in Pertmaster we can open P3/PE files directly without any problems too. In PE we can move some effects of MC analysis back to PE.

Today MC for P3 is history if we must to abandon old P3! Pertmaster is good tool, not only for MC analysis!!!! but british producer of Pertmaster is interested only in MC analysis..... It is big problem from my point of view, but You are interested only in MC :-( Meybe my destription is not valuable for You, Dear Scott, but in my oppinion scheduling and MC in one system is better than slowly action of two separated systems so comparrision of this two MC systems is not simply.

good luck for Planning Planet users :-)
Tomek from Poland
Scott Tate
User offline. Last seen 10 years 44 weeks ago. Offline
Joined: 4 Mar 2005
Posts: 28
Groups: None
Are there any Pertmaster users who can answer the questions I posted below when inquiring about Monte Carlo?
Dayanidhi Dhandapany
User offline. Last seen 1 year 37 weeks ago. Offline
Joined: 18 Mar 2003
Posts: 472
Groups: None
Can someone send me a copy of evaluation version of Monte Carlo software that supports P3 version 3.0 or 3.1 to my email id plannerdaya@yahoo.com

TQ

Daya
Scott Tate
User offline. Last seen 10 years 44 weeks ago. Offline
Joined: 4 Mar 2005
Posts: 28
Groups: None
Can someone provide a brief summary of the benefits and limitations of using Monte Carlo? For example, is there a maximum number of activities the schedule can have or does a detailed schedule have to be done at a summary level in order to meet the activity number limitation? Also, can hammocks be created from the detailed level to make summary bars and loaded into Monte Carlo? Does Monte Carlo recognize hammocks or does it require logic as well? Is there a difference between the way Monte Carlo reads P3Win versus P3e/c? Thanks for any information anyone has on this topic.
osama yassin
User offline. Last seen 14 years 9 weeks ago. Offline
Joined: 6 Dec 2004
Posts: 11
Groups: None
[please do not post the same message in multiple threads. Thank you.]
Alex Wong
User offline. Last seen 6 years 21 weeks ago. Offline
Joined: 12 Feb 2003
Posts: 874
Groups: TILOS
Dear All,

From reading the message I think we focus on the tools itself but not the process. It is far more important to have a systemic approach to address risk and risk management.

Yes part of your schedule should allow certain contingency for risk elements what I call mitigation. **the four categories of risk treatment (Avoid, Transfer, Mitigate and Accept) However, this is only part of the risk management. I think it is very important for the project planner understand the full extend of the risk management. And PMBOK is a good start.

Alex
Robert Foedisch
User offline. Last seen 8 years 12 weeks ago. Offline
Joined: 7 Nov 2004
Posts: 34
Groups: None
I Would think it would depend on what your analyzing
Kris Harry
User offline. Last seen 7 years 51 weeks ago. Offline
Joined: 4 Oct 2004
Posts: 47
Groups: None
What is normally the scopr of schedule risk analysis - does it take care of time, cost etc
Robert Foedisch
User offline. Last seen 8 years 12 weeks ago. Offline
Joined: 7 Nov 2004
Posts: 34
Groups: None
I have studied risk analysis and would like to help. Can someone describe the scenario?
Bob
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Daya,

You are probably expert in practice of risk but I am able to explain You semi-theoretical problem/question. From theoretical point of view uncertainty it is not the same with risk (aspect of classification). If we know probability distribution we have risk. In opposite case we have uncertainty but in practice we have not exact knowledge abut distribution, of course BUT we still using MC simmulation ... "theory is not near to practice" :-))

If we want to model cofussing/risky inspection we can to create discret event like an probable task with probable existence. Additional aspect is risky duration inspection AND risky consequences of inspection :-) but it is soul of simulation. We may to analyse it without big problem so MCS is most prescribed method in our :-) industry (meybe decision tree too, but we may to make schedules not trees in practice). Schedule risk it is solid knowlegde like in sentence "in practice good theory is the best".

Good Luck
Tomek from Poland ;-)
P.S. Happy new year on the world, all world - without radical uncertainty ;-)
Tomasz Wiatr
User offline. Last seen 1 year 4 weeks ago. Offline
Joined: 22 Dec 2004
Posts: 35
Groups: None
Dear Pramesh,

You are starter of this discussion so I would like to write You :-) In my oppinion risk analysis with Excel is simple but distant from ideal (waste of secondary analysis of MC simulation effects - we need @Risk or CrystalBall). I am interested in Project Risk so I work with it and I have deep view.

In my oppinion most friendly AND most advanced is Pertmaster! I am user of it in my scientific work because in practice (!) advanced risk analysis is not common desirable in construction (it is only for connoisseurs :-) Today problem is too high price of Pertmaster :-( but other systems are expensive too. It is second problem of using of MCS in "real live projects" (your citation :-)).

We may to discuss about MCS in AEC industry. For me MCS was started with aid of Pertmaster, not with basic Excel. You can fall in love MCS with Pertmaster, too. Pertmaster is easier than Excel in aspect of MCS !!! I am good modeller of risk and modeller of complex projects. In my oppinion Pertmaster must to have own topic on the our discussion list, not only like Risk Analaysis tool (not only!).

Good Luck
Dayanidhi Dhandapany
User offline. Last seen 1 year 37 weeks ago. Offline
Joined: 18 Mar 2003
Posts: 472
Groups: None
Dear Vladimir,

Thanks for your information, now i’ve some clear picture about risk analysis.


Best Regards

Daya
I will describe approaches that are used here and supported by Spider Project.

All project information (durations, volumes of work to be done, costs, material requirements) is collected with three estimates - optimistic, most probable and pessimistic.

Potential risk events (like third-party inspections, weather events, delivery delays etc.) are estimated (probability and impact). Impact can lead to additional work that shall be done, delays, additional cost, etc.
You shall choose which risk events should be considered (PMBOK Guide described this approach). Let’s call these risk events as red risks.

Red risks with the probability 90% or more (almost certain) will be included in the optimistic project scenario, 50% or more - in the probable scenario and all red risks - in the pessimistic scenario.
Optimistic version is based on the optimistic estimates of the project parameters, probable - on most probable and pessimistic - on pessimistic.

Three schedules are calculated. As the result we receive optimistic, most probable and pessimistic estimates of project duration and cost (the same about project phases). The next task - to estimate what are the desired probabilities of successful achievement of the project targets. If you set these probabilities the software will calculate desired project (phase) finish date and project (phase) budget. And vise versa - setting project targets you will receive the probability of meeting these targets. It is calculated reconstructing the probability curves. These curves remind beta distribution but are sharper depending on the number of project activities in the resource critical path (for schedule) and the total number of project activities (for budget). If you do it manually define the curve yourself (PERT is one of the options). You may be wrong and your estimate will be for example 10% more or less than actual. What is much more important - this error will be the same (or close) in the future estimates. Later you shall control the trends of these probabilities and estimate project performance - if the probability of meeting project targets is rising then the performance was successful.

So you will have target dates and costs and 3 schedules. Use optimistic schedule to set tasks for the project implementors, target dates will define project management team targets, contract targets should include additional reserves (or profit for costs).

Managing the project you shall update all schedules to be able to estimate current probability of meeting all project targets (example - team target finish probability 70%, contract target finish probability - 85%).

Contract schedule is high level and the result of negotiations. Usually you don;t need to show your internal detailed work plans.
Dayanidhi Dhandapany
User offline. Last seen 1 year 37 weeks ago. Offline
Joined: 18 Mar 2003
Posts: 472
Groups: None
I’ve a little exposure in risk-management. If someone clearly explain me the following i would be delighted.

When we develop a construction schedule, what are all the key elements to be considered as risk?. How to show these elements in a construction schedule?. Is it just by increasing 10~20% of Original duration in order to take care of some predicted risk?.

Is there any relationship between Risk and Uncertainty?.

In EPC projects, there is lot of uncertainity involved(third-party inspection etc before shipment) in delivery of procurement materials, How to justify these things in the schedule, whether client can be convinced of showing additional duration to take care of this delivery as a separate activity in the schedule development stage?.

Which is the best prescribed method of risk analysis in the construction industry?.

Regards

Daya

The number of necessary iterations create a huge problem with using Monte Carlo simulation. It is useless in large projects. You can receive rough approximation but you will not be able to rely on results trends. We always insist that probability trends matters much more that probabilities themselves. You can use rough approximation if the error is the same in each following simulation.
Forum Guest
User offline. Last seen 9 years 16 weeks ago. Offline
Joined: 28 Jan 2009
Posts: 0
Groups: None
With Excel>?

I am currently assist on a Motorway project located in India.

I used MathCad in simulate the probabality of risk in productivity rate of each single activity. These to define the Value at Risk of such phases.

Simple & easy way(as per your query >?
It would save a huge amount of time using MatCad in comparison to Excel.

Mohd Razif
Planner/Scheduler P3.1
Kuala Lumpur
Malaysia
rahil khan
User offline. Last seen 15 years 5 weeks ago. Offline
Joined: 10 May 2004
Posts: 7
first of all there are many methods by which risks can be analysed, like the 3-point estimate method or the Clean sheet method.

as long as u want to keep it simple, i think excel could be used, but simple method means less reliable.
MK TSE
User offline. Last seen 30 weeks 5 days ago. Offline
Joined: 27 Feb 2002
Posts: 552
Groups: None
There are 2 type of analysis, Quality and Quantity and both can be applied to a programme. For Monte Carlo or function in Open Plan, they are classified into Quality type. For Quantity type, either use some software tools or build your own in Excel. However, use database is preferable since most quantitive analysis tool are established on database.
Rafael Davila
User offline. Last seen 6 hours 4 min ago. Offline
Joined: 1 Mar 2004
Posts: 4779
Be carefull with simplistic PERT risk analysis. It is usually wrong as the individual risk distribution of near critical activities might affect the resulting distribution.

As far as I know there is no simple alternative other than a simulation.

If all your activieties are in tandem, none in parallel then you might use a simplistic approach.
Guy Hindley
User offline. Last seen 22 weeks 6 days ago. Offline
Joined: 28 Sep 2001
Posts: 91
Groups: None
I have carried out schedule risk analysis, but using proprietary tools as distinct from Excel. Tools I have used have evolved over the years. I have had experience of Primavera Monte Carlo, but currently mainly use the schedule risk functionality of Open Plan. Colleagues I work with also use other tools.

Would this sort of information help you, or are you only looking to use Excel?