Monte Carlo Simulation Convergence

September 14, 2015 Q: Monte Carlo Simulation Convergence I defines maximum number of simulations 2000, however simulations stop before it. I believe it is because of convergence. How does it work? How can I correctly define convergence criteria? Do you calculate convergence based on cost? A: Convergence monitoring allows to stop Monte Carlo simulations when additional iterations don't contribute to improve accuracy of analysis. Please remember that in most cases we don't define uncertainties very accurately. For example, task duration can be between 5 and 7 days. However it is possible that task duration can be between 4.5 and 7.5 days. In many cases we don't have reliable historical data to definitely say that low duration will be 4.5 or 5 days. In other words we define uncertainties with 10% accuracy. Therefore it does not make sense to perform very accurate Monte Carlo simulations with accuracy 0.1%. Convergence parameters are defined in Project Options Dialog, Calculation tab (click on Schedule -> Project Options): To enable convergence you may just click on "Stop simulation when converged". If this box is not checked, number of Monte Carlo simulations will be exactly as shown above this check box (Maximum number of simulations). Default Monte Carlo convergence parameters are usually applicable to most schedules, so you don't need to change anything. RiskyProject calculates mean and standard deviation of project duration on each iteration. The different between mean (and standard deviation) calculated on current and previous iteration should be no greater than certain number. The default value is 0.15%. This criteria should be met on certain number of iterations. The default value is 20 iterations. If within 20 iterations between mean (and standard deviation) calculated on current and previous iteration is greater than 0.15%, RiskyProject would check this criteria over next 20 iterations. RiskyProject tests convergence criteria only for duration, not for cost. "See intaver's forum for more details". http://www.intaver.com/IntaverFrm/viewtopic.php?f=2&t=1122

Relative vs. fixed (absolute) delays for project risks

September 2, 2015 Q: Relative vs. fixed (absolute) delays for project risks When you are adding risks, their is a choice between fixed (absolute) durations in days or relative delays. When should you use fixed rather than relative outcomes? MTC465 Posts: 3Joined: Tue Jun 06, 2006 12:57 amLocation: USA A: When you perform quantitative project risk analysis in RiskyProject, you need to define risks with their outcomes or how the risk will affect particular task or resource. Risks belonging to different risk categories will have different outcomes. Risk affecting task duration may cause fixed delay (certain number of days), relative delay (for example task may be delayed on 10%), task restart, and task cancelation. You may also enter combination of these outcomes, as it shown on the screen shot below. They are called alternatives. Particular type of outcome depends on your schedule and data you have available. For example, you know that the risk “component is delayed on customs” is a fixed delay, because it takes fixed number of days to clear the customs. However risk “delay caused by contractor” is a relative delay because it depends on duration of original work: contractor just works slow on underestimated duration of the task. In certain cases, for example due to changing of requirements, the task must be started again. In this case the outcome should be “restart task”. Please note that RiskyProject also have fixed and relative increase of cost and income. Also if you include negative values to duration, cost, and income increase, it will be opportunity, not a threat. If some risk alternatives have positive outcome and some negative, the risk will have both threat and opportunity. http://www.intaver.com/IntaverFrm/viewtopic.php?f=3&t=1102

Uncertainty to non-critical tasks for portfolio analysis

September 9, 2015 Q: Uncertainty to non-critical tasks for portfolio analysis Do I need to add uncertainty to non-critical tasks? Typically we need to run Monte Carlo schedule risk analysis on large projects. In the past, for simplicity, we have only assigned uncertainties to activities on the critical path. I saw a presentation recently that advised that we need to assign uncertainties to all tasks. Is the correct and what are the implications to our analysis if we don’t. A: Short answer is yes, best practice is to add uncertainty to all your project activities when performing Monte Carlo simulations. In reality this could be further refined by add “that can vary in duration” to this advice. Obviously deadlines fall into this category, but there are also some routine activities of very short length (~ 1 day) that can be safely ignored. However on deterministic schedules only a relatively few activities will be critical. If you do not add uncertainty to the other tasks, the Monte Carlo schedule analysis will not capture the impact of near critical activities, merge bias, and the portfolio effect. In probabilistic project analysis, there can be multiple critical paths where near critical activities can become critical. The Crucial task view is a visual representation of how often activities will land on the critical path. In addition, if you only add uncertainty to critical activities, this may cause errors in the sensitivity analysis. http://www.intaver.com/IntaverFrm/viewtopic.php?f=6&t=1116

Probabilistic Cash Flow of Project Portfolio

September 4, 2015 Q: Probabilistic Cash Flow of Project Portfolio Can I present a probabilistic cash flow chart as a result of Monte Carlo simulations not for individual projects, but for complete project portfolio? A: Probabilistic cash flow chart is generated for individual project only. Cash flow analysis is performed as part of project's Monte Carlo simulation. The issue is uncertainties in each project can be defined differently and it can lead to incorrect results, when we summarize them for project portfolio. However we are currently working on new version of the software, which will define common uncertainty "bands" for project portfolio. If these "bands" are shared for all project within a portfolio, it is possible to define standardized set of uncertainties for project portfolio. Just to remind you, RiskyProject performs Monte Carlo simulation of project schedule and generate statistical distribution of fixed and variable costs on each time interval. In can be done weekly, monthly, by-monthly, etc. The results of each project is presented in form of chart: Dark blue color of each bar present low estimate of post for the given period of time, lighter color will present base cost, and very light color will present low cost. You can also be presented with cash flow of current schedule (red line). It is planned (original) cash flow. Actual cash flow can be shown as a green line, if the project has actuals. You can click on each chart to see detailed statistical distribution for cost of each time interval. http://www.intaver.com/IntaverFrm/viewtopic.php?f=6&t=1106

Switch from project to portfolio risk analysis

September 1, 2015 Q: Switch from project to portfolio risk analysis I use RiskyProject professional for project risk analysis. I would like to do portfolio risk analysis using RiskyProject Enterprise. How can I switch from project to portfolio risk analysis? A: RiskyProject Lite and RiskyProject Professional perform analysis of single project only. If you want to preform risk analysis based on shared list of risks for your organization, you would need to use RiskyProject Enterprise. RiskyProject Enterprise you two components: server and client. The server component is a database which needs to be installed on the server. The client component is essentially RiskyProject Professional or RiskyProject Lite, which can be installed on client computer. The client component needs to be configured to talk with the database on the server. If you already have RiskyProject Professional you need to install server component and configure RiskyProject Professional connection to the database. All instructions how to install server component can be found in RiskyProject Administrator’s guide, which comes with RiskyProject install package. Server database is using Microsoft SQL Server or SQL Server Express version 2005 and up. When you install RiskyProject server database you need to go Tools->Update RiskyProject License. The License Wizard dialog will come up. There are three possible configurations of RiskyProject: Trial License, Regular Desktop license, and Enterprise client. You current configuration will be grayed out. You select different configuration: in your case it will be Enterprise client. After this you need to close RiskyProject Professional. When you start RiskyProject again you need to configure connection to the database as it is described in RiskyProject Administrator’s guide. After you configure connection between RiskyProject Professional and server database you may perform Monte Carlo Schedule and Cost risk analysis for all particular projects within a portfolio. Risks and results of Monte Carlo simulations will be saved in the database and shared between different users on your organization. http://www.intaver.com/IntaverFrm/viewtopic.php?f=6&t=1101

How do I show Project Summary in the Statistics Report?

September 4, 2015 Q: How do I show Project Summary in the Statistics Report? I s there any way to have or show the project results as shown in the Project Summary in the Statistics Report? Project results should include results of Monte Carlo simulations: statistical summary, histograms and percentiles for project duration, cost, start time, finish time, etc. A: The Statistics Report is designed to show the results of your Monte Carlo simulations for any activities in your project, including summary activities. Using the Statistics Report you can show Histograms, Cumulative Probability Plots for Cost, Finish Time, Duration, and Start Time. You can also see Sensitivity or Tornado charts for Finish Times (that is which activities in the precedent network have the most potential to impact the finish time of the task). The Statistics Report does not support either analysis of Work and the Cost vs Duration Scatter Plot (Joint Confidence). To use the Statistics Report to generate the charts for the Project requires that you insert a Summary Activity at Level One of the highest level of the schedule. This can be part of your project schedule or can be inserted specifically for reporting purposes. Once this Project Summary activity is inserted the summary results for your Monte Carlo risk analysis will be available in the Statistics Report. http://www.intaver.com/IntaverFrm/viewtopic.php?f=3&t=1105

Status of Monte Carlo simulation

September 6, 2015 Q: Status of Monte Carlo simulation When I click the Calculate button to run the Monte Carlo simulation, I no longer see a status as in previous versions. Just downloaded version 6.1 to check out the new features. RiskyProject seems to be calculating, as I can see results in the Analysis view, but there is no progress bar. Expert Posts: 3Joined: Sun Jul 09, 2006 10:11 pm A: With version 6 and then with version 6.1, there have been significant increases in the speed of Monte Carlo simulation calculations. Especially with version 6.1 where we have added multiprocessing if you have a relatively new computer, depending upon the project data (number of tasks, resources, costs, risks etc.) time to run the project schedule and cost risk analysis has been reduced from 10-30 times. In the cases of smaller project (<1000 activities) with no or small numbers of resources and risks the simulation occurs so quickly now that the progress bar does not appear. This is currently an undocumented feature dubbed “Instant Monte Carlo”. http://www.intaver.com/IntaverFrm/viewtopic.php?f=3&t=1109