Event Chains Methodology in Project Management

Posted: Sunday, October 09, 2005

by Ken McKinley
Intaver Institute Inc.

Ken McKinley

align=center>Intaver Institute Inc.

303, 6707, Elbow Drive S.W.

Calgary, AB, T2V0E5, Canada

Phone: 1(403)6922252

Fax: 1(403)2594533

Email: kmckinley@intaver.com

www.intaver.com

Estimations in Project Management

You created a well-balanced project schedule and thought that you had taken into account almost every possible scenario and risk. However, as soon as you started implementing your project plan, something happened and your schedule became obsolete. This “something" is an unpredictable event. As a result, you have either to significantly update or create a new project schedule and then, another unpredictable event occurs. This scenario is very common for projects with multiple risks and uncertainties and especially true in research and development projects such as those found in the software industry. Should we completely give up scheduling, risk management, and concentrate only on high-level project planning, or is there still a way to provide realistic estimates for project schedules that have multiple uncertainties?

We can perform estimations related to epistemic (knowledge driven) uncertainties by analyzing historical data and by tracking the current project’s performance. The problem is both methods cannot change the subjective nature of epistemic uncertainties. Analysis of historical data is subjective and negatively affected by the psychological heuristics and biases. What would happen if you kept accurate records? The answer depends on what type of tasks you are trying to estimate. In some industries, such as construction and manufacturing, these records are available.

However, in many other industries, especially research and development projects, significant number of tasks have never been done before therefore, historical records may not be available or very useful. Very often a similar, but not exact, task has been done before. Can you use this information about previous tasks as an analog for the estimation? Another problem with historical data is that if there was a problem with the activity before, project managers will avoid making the same mistake again.

Because of these problems with historical data, the tracking of actual project performance remains one of the primary means of keeping projects on track. The goal is that by tracking actual performance, we can somehow reduce uncertainties during the course of an activity and derive better estimates of duration and cost. However, the problem of estimation remains for the reminder of the activity and project.

Therefore, because we recognize that it is difficult to determine a single number associated with task duration and cost, the current practice is to overcome this deficiency by defining a range of numbers or a statistical distribution associated with this range for cost and duration. For example, the range for a task can be from 4 and 7 days. However, if historical records are unavailable, we will still have the same problem. These estimates will be as subjective as if they were defined by a single number. If the range estimations are as subjective as a single number estimate, then analysis by using ‘classic’ Monte Carlo simulation may not provide estimates that are any more accurate than deterministic project schedules.

Overview of Event Chain Methodology

Therefore, we are drawn to the conclusion that if uncertainties are expressed as events with outcomes, it will significantly simplify our project management estimations. By mitigating some biases in estimation, we can develop numbers that are more accurate for task duration, cost, and other project parameters. Once we have this data, we can perform quantitative analysis and determine how uncertainties in each particular task will affect the main project parameters: project duration, cost, finish time, and success rate. However, real projects are very complex they have multiple risks that have the potential to trigger other risks. Risks can have different outcomes in one scenario a risk will delay a task, in another scenario the same risk will cancel it. In addition, some risks are correlated with each other. Therefore, the problem remains how to model these complex processes so that it becomes practical for project management.

Event Chain Methodology proposes to solve this problem. It is important to note that Event Chain Methodology is not a simulation or risk analysis method. It is based on existing analysis methodologies including Monte Carlo simulation, Bayesian approach and others. Event Chain Methodology is a method of modeling of uncertainties for different time-related business and technological processes including project management.

Event Chain Methodology is based on six major principles.

An activity (task) in most real life processes is not a continuous uniform procedure. It is affected by external events, which transform an activity from one state to another. It is important to point out that these events occur during the course of an activity. The moment, when an event occurs, in most cases is probabilistic and we can define it using statistical distribution. Events (risks) can have a negative impact on the project. For example, the event “delayed arrival of component" can cause a delay in an activity. However, the opposite is also true, events can positively affect an activity, e.g. reduce costs.
Events can cause other events, which will create event chains. These event chains can significantly affect the course of the project. For example, requirement changes can cause a delay of a task. To accelerate the activity, a resource is allocated from another activity which can lead to a missed deadline. Eventually, this can lead to the failure of the project. Events may instantly trigger other events or transform an activity to another state. The notion of state is very important as states can serve as a precondition for other events. For example, if a change of requirements causes a delay, it transforms the activity to a different state. In this state, the event “reallocate resource" can occur. Alternatively, it is possible, if the task is in certain state, an event cannot occur.
Once events and event chains are defined, we can perform quantitative analysis using Monte Carlo simulation to determine uncertainties and quantify the cumulative impact of the events. Sometimes we can supplement information about uncertainties expressed as an event with distributions related to duration for start time, cost, and other parameters, as done in classic Monte Carlo simulations. However, in these cases it is important to discriminate between the factors that are contributing to the distribution and the results of events to avoid a double count of the same factors.
The event chains that have the most potential to affect the projects are the “critical chains of events." By identifying critical chains of events, we can mitigate their negative effects. We can identify these critical chains of events by analyzing the correlations between main the project parameters, such as project duration or cost, and the event chains. Events or event chains can be displayed using a tornado diagram where critical event chains are shown on the top.
Probabilities and impact of the events are obtained from the historical data. Monitoring the activity's progress ensures we use updated information to perform the analysis. In many projects, it is hard to determine which historical data we should use as an analog for future analysis. For example in most cases, in research and development, new projects differ from the previous projects. We can accomplish the proper selection of analogs for the historical data by applying analysis using a Bayesian approach. In addition, during the course of the project, we can recalculate the probability and time of the events based on actual data.
Event Chain Diagrams are visualizations that show the relationships between events and tasks and how the events affect each other. By using Event Chain Diagrams to visualize events and event chains, we can simplify the modeling and analysis of risks and uncertainties.

Event Chain Methodology Phenomena

The application of Event Chain Methodology can lead to some interesting phenomena. Here are some examples:

Sometimes events can cause the start of an activity that has already been completed. This is a very common scenario for real life projects sometimes a previous activity must be repeated based on the results of a succeeding activity. Modeling of these scenarios using event chain methodology is very simple. We do not have to update the original project schedule, we just need to create an event and assign it to an activity that points to the previous activity. In addition, we need to define a limit to the number of times activity can be repeated.
Events can generate other activities that are not in the original project schedule. These are activities related to the mitigation plan. They are modeled outside of original project schedule and assigned to the event. The original schedule is augmented with these activities when the event occurs.
One potential event is the reassignment of a resource from one activity to another, which can occur under certain conditions. For example, if an activity requires more resources to complete it within a fixed period, this will trigger an event to reallocate the resource from another activity. Reallocation of resources can also occur when activity duration reaches a certain deadline or the cost exceeds a certain value. Events can be used to model different situations with resources, e.g. temporary leave, illness, vacations, etc. In some cases this can create an event chain: due to an illness, a resource from another activity would be borrowed to accomplish a specific task.
Events can cause other events to occur either immediately or with a delay. The delay is a property of the event. The delay can be deterministic, but in most cases, it is probabilistic. If we know the time of the original event and the delay, it is possible to determine when the new event can happen and in some cases, the activity that will be associated with it.

Conclusions

The beauty of this approach is that it is includes a very well defined mathematical model that can be easily implemented as a software algorithm. Project managers must define project schedules and risk lists or risk breakdown structures. For each risk, the manager defines the chance the risk will occur, the risk’s impact (delay, increase cost, trigger other risks, cancel task, etc.), and when will the risk occur during the course of activity.

Event Chain Methodology allows us to model projects with uncertainties in a much simpler manner. It also allows us to mitigate psychological biases related estimation and as a result provide better forecasts and project tracking. If risk and uncertainties based on Event Chain Methodology are defined properly, your project schedule should be much more robust. Remember, most project managers actively create and update project schedules and risk lists. Event chain methodology allows you to combine both lists to provide a simple answer to the central question of project management - how long will the project take and how much will it cost if an event occurs.

This Article has been viewed 350 times. (Not updated in real-time.)

No comments yet.

We want your comments! If you can read this, you don't have javascript enabled, so you can't use this comment system. Please enable javascript.

Viewed from Cache.
Load time: 0.000 seconds.