The Project Manager's Corner: Scheduling TMT I
Gary Sanders, TMT Project Manager

February 2007

Last month I discussed adjustments to the design of TMT and to its estimated cost. I described our progress into the last phase of TMT design. By May 2008 we plan to complete the site-independent Preliminary Design of TMT. This means that we will have fully defined the design that we intend to build and that this design can be shown to meet the design requirements. Following that phase, we will have selected the TMT site and will proceed to complete site-specific elements of the design, such as different foundations for different local geology. Final design, where detailed construction documents are completed for industry, follows and this initiates construction.

One of the important elements of this process is defining the complete schedule to construct TMT. How does a big project like TMT go about this?

The simplest schedules are an ordered list of tasks with each assigned an estimated time to complete. Such schedules are commonly shown as a list or a chart of time bars on a graph that looks like a waterfall of bars scrolling down the list and marching out in time. The figure displays such a schedule as a cartoon. This is a good way to start organizing early thoughts. But it is far too simplistic for a big project.

Some tasks must wait till other tasks are fully completed. Some may progress in parallel. Some may wait for several other tasks to be completed. Some represent parts of the project that if delayed will delay the entire project. These are called critical tasks. Others can take place whenever there is time for them within broad limits. Delaying these does not stretch out the entire project. Some tasks can be made to progress more quickly than planned, for example by using more shifts per day, or weekends, for production tasks. Others depend upon a unique resource or method and cannot be hastened. Buying and receiving a large ingot of a specialty glass may simply not be speeded if there is only one supplier with one oven and one process able to meet the requirement.

A project like TMT may have thousands or tens of thousands of tasks in the final schedule. What do you need to design such a schedule?

First, the plan must be made for each major subsystem by experienced designers of such systems. These designers must have real technical expertise for the tasks planned and they must have actually carried out similar projects before. This enables realism in the schedule design and the strategy. It is best when these experts are the actual team members responsible for carrying out the schedule. They will then be working to their own plan. This is schedule ownership.

The second required element is hinted in my reference to each subsystem. This implies an organization of the work into parallel, meaningful subprojects that come together into the entire project. Each of these is a delivered item. Delivering all of these items leads to delivering the entire project. This way of organizing the entire work of a project into all deliverable items and tasks is called a Work Breakdown Structure or WBS. The WBS defines how the cost estimate is tabulated, how the schedule is made up of tasks and how the work will be managed.

The third required element is the overall architecture of the work plan and any imposed constraints. Tasks cannot begin before funds are available, for example. The project may be required to be completed before some external milestone date. This architecture and top level set of milestones are adopted as guidance at the beginning of the schedule planning.

Next month I will describe the remaining steps in building the typical large project schedule and discuss our vision of how TMT will be implemented in time.