Коллеги, а у кого есть опыт общения/пользования XER Toolkit и Safran Risk? Как оно в сравнении с Primavera risk analysis?

Присоеденяйтесь к официальной группе ИСАР в Facebook

Обучение для риск менеджеров


  1. Вот что по этому поводу думают мои инностранные друзья. Вопрос прочитали 3000 человек. С создателями / представителями Tamara, Safran могу познакомить. XER я так понял лажал, никто не пользуется.

  2. Alex Sidorenko спасибо за организованную дискуссию, здорово когда есть возможность собрать мнение такой широкой аудитории! Tamara – надо обратить внимание, интересно!

  3. It’s interesting that people praise the Safran Risk ‘Sensitivity by Exclusion’. The method is technically wrong, and analogous to a very old sensitivity method used in Monte Carlo simulation.
    It’s wrong because:
    (1) the delivery time (and cost) distribution is driven by interactions between uncertainties and risks. Removing one risk changes the interaction, so one does not see compounding effects;
    (2) One would have to remove uncertainty on each task duration too for consistency – the main driver may just be task duration uncertainty, not risk events
    (3) The main driver could also be some productivity factor, or factors
    (4) On a large schedule (say 20,000 tasks with 200 risks), Safran would take forever to perform the analysis.

  4. As a schedule is just someones idea on how the project will be executed and almost certainly not the way it will be executed, I always treat the output with the some care and the input. The more detailed the network the more likely the sequencing of the activities will be different. Like Steve Grey and David Hulett, I prefer to work with smaller networks and think how best to model the main strategic risks. The 5000 activity guys then say you are missing nodal bias. Maybe , but they are down in the weeds and missing the bigger picture. I chuckled when one of the comments praised the ability to work in minutes!! I support turnarounds and yes you need to work in hours as the tasks can be short, but if your tasks are in minutes then the detail is way too much. Many activities sit in FS chains, so just model them as one activity.

    I also believe @Risk for Project can model time dependent costs. I built a cost template that imports the schedule. That said, like you, I find @Risk, is buggy. The passing of data between Project and Excel and @Risk is so complex. My models used to hang for 20 minutes, but they wrote me a patch. I use the uncertainty and risk driver approach and make little use of correlations. Vose showed @Risk where there correlation modelling was flawed, but the “correct” way is not easy to implement. I believe this is more black art than science.

Comments are closed.