Critical path analysis

Critical path analysis (CPA) was developed in the days when projects were seen as largely sequential in nature. The critical path could be calculated manually by any competent project manager even on quite large projects.

Today, the demands on the technique and its practitioners have changed radically. Most projects are fast-track, which means that there is little, if any, float. Unfortunately, many people using CPA are not fully trained or have little experience in how to use it for complex projects. They rely on software to calculate the critical path, but, as with any software, if the information input is inaccurate or inadequate it will create an incorrect critical path.

What is the critical path?

The critical path is the longest uninterrupted sequence of linked bars (in relation to the project timeline) stretching back from the last scheduled task. The test for criticality is simple: if a task is delayed or takes longer and as a consequence puts the project end date back, then it is critical. For non-critical tasks, it works out the float (potential slippage) that tasks can be delayed or extended by without affecting the project end date.

CPA is based on task durations and therefore does not necessarily embrace every aspect of a project. Projects today are vulnerable to a lack of professionally skilled resources, risk factors, lack of attention to accurate estimating, the over-zealous demands and changes made by clients, and cost overruns.

You can estimate the expected length of tasks, but if you are not able to predict a resource shortage, you could have a situation where the initial plan shows tasks as non-critical with float, but due to those shortages the task will overrun and possibly put back the project end date. So CPA in itself is no guarantee of success.

The question is: how do you sensibly estimate task durations for scheduling purposes? Should you deliberately include contingency in the estimates for activities?

Estimating

Companies tender to win contracts. When you start to produce the contract plan, you may find that the timeframe, skilled labour issues and the budget is much more challenging in reality.

In many cases, tender programmes are not detailed enough or properly thought through and so the best way to approach the planning process for a live project is from scratch. Create a list of tasks and best duration estimates. Then, without reference to any time constraints, link the tasks to produce a viable chronological plan. Next, taking into account the project timeframe and any major milestones, refine the task durations and link types to meet the completion date for the project.

That demand to refine the plan will in practice mean negotiating with subcontractors and internal management. The greater the in-depth knowledge of how the project fits together, the easier it is to understand the crunch points in the project. For instance, if there are time periods in the project where there is little or no float on any tasks, then you know that careful and precise management is needed at these points.

Linking

In today’s projects, many tasks are performed in parallel and therefore the practice of linking is vital to schedule projects correctly.

Since many software packages are now so graphical, most people (even planners) draw the task bars direct on the chart, often using a mouse to make the link. Even though many packages have a network diagram (a workflow representation of tasks as boxes that show the order in which tasks are performed and how one task is linked to another), these are becoming redundant. However, for most software packages, it is hard to represent parts of tasks overlapping each other .

Modern projects, due to time constraints, need to accommodate many overlapping tasks and if you require a more accurate critical path, then more construction-friendly products allow you to draw multiple bars on a line and link them separately.

This is more time-consuming to input, but is more precise for scheduling and it means any critical path can then just run through only certain parts (bars) of a task.

Linking tasks can be onerous, so more graphical software products that make maximum use of the mouse are essential. They allow for all types of links to be drawn with the mouse and will calculate lags automatically from their chart position. Some software products allow mid-point linking of bars, which many users prefer.

Updating the Critical Path

The critical path is only a snapshot at a point in time. Many people produce a critical path at the start of a job and only update the plan by adding percentage complete progress figures. While this is fine for reporting progress against plan to the client, as the project moves on, the actual and future estimated task start and end dates and durations will change. If you are still attempting to control the project using the original plan, the critical path and the schedule will be out of date.

Many project managers often create two versions of the project: one with the original plan and original critical path with progress percentages shown; another where two bars per task line are shown – one for current plan and one underneath for the original plan. This allows you to update the current plan for actuals and forecasts and so the critical path can be recalculated at each progress update point.

Is CPA time consuming?

The simple answer is yes. A fully linked bar chart can take two to five times longer to create than a chart simply drawn with no linking. The time is taken up thinking through clearly the most appropriate linking type between task bars. There are four types of links: finish to start (FS); start to start (SS); finish to finish (FF) and start to finish(SF). If you do not pick the correct one, when the chart is updated to show actual progress, the outstanding tasks will not be scheduled correctly.

For instance, the original linking type for the painting tasks might be correct. However, if the first coat takes 16 days and the user forgets to update the duration for the second coat, then you would have an illogical situation where the second coat would end before the first coat finished.

Construction-friendly software allows you make links at the end of each pair of bars (known as hammocks) where the duration of the second bar is determined automatically by the other one .

Hammocks can also be used to schedule tasks where the duration of bar fits into the time available between two tasks. Not all project software allows you to do this.

is the Critical Path is correct?

You cannot prove a critical path if you do not link bars and show the logic, but even today some people believe they can highlight a critical path visually. Even though that may be possible, it is usually a risky assumption, as CPA is a highly complex algorithm and depends on the link types, durations and task constraints (such as a fixed end date for a certain task).

There are a number of common pitfalls that inexperienced persons encounter. Here are a few guidelines:

All bars on the critical path must have at least one link going into it and one link leaving it unless it is at the start or end of a path. Link breaks in the middle of a project will artificially cause large float values in the early tasks and would lead people falsely to believe delays were possible without harming the project.

Many people make the mistake of defining links between tasks as one-to-one links. In most cases there are multiple relationships between tasks. It is not unusual in modern projects that the number of links can exceed the number of tasks by a factor of up to four.

In projects with tight timescales, if tasks are linked correctly, it is not unusual for as many as 75% or more of the tasks to be critical.

In complex projects, there is little float and the critical path will often extend from the first scheduled task to the last scheduled task.

There are people who question the use of CPA as a management tool for modern projects. But project managers within client organisations increasingly demand it. In retail or customer service buildings, delays can have major implications for a company and project managers are insisting on more and more information to feed their project control systems to avoid these costly delays.