CORE PRACTICE: GLOBAL OPTIMIZATION, NOT LOCAL
The main pitfall to avoid with real-time scheduling is to continue with local optimization that was necessary with static plans. The results you get with real-time scheduling are superior when you use it for global rather than local optimization.
Static plans are not useful in execution. Resources can only work on what’s available, not what was planned. That’s why, with static plans, day-to-day execution is managed with local optimization metrics like resource utilization and efficiencies. However, there is a big difference between keeping resources busy and getting projects done (more projects faster).
Let’s conduct the following thought experiment to understand the difference between local and global optimization.
There are three identical work-streams in this experiment, and each work-stream is comprised of three main tasks:
- Blue Tasks (B1, B2 and B3) that are performed by the Blue resources;
- Red Tasks (R1, R2 and R3) that are performed by the Red resources; and
- Gloden Tasks (G1, G2 and G3) that are performed by the Golden resources.
Now, imagine that you can execute these three work-streams in one of the following two modes:
- LOCAL OPTIMIZATION MODE: “Maximize Resource Utilization” whereby resources are fully utilized and all the three work-streams are executed in parallel.
- GLOBAL OPTIMIZATION MODE: “Maximize Project Velocity and Completions”, whereby all the required resources are assigned to one task at a time and only after finishing that task can they start another. For example, Blue resources will have to finish Task B1 before they can start Task B2. Resources experience some idle time in this mode.
Let’s further assume that:
- In the “Global Optimization” mode, Blue resources can finish one Blue Task in 6 weeks, Red resources can finish one Red Task in 10 weeks, and Golden resources can finish one Golden Task in 4 weeks. This mode requires 18 (6x3) weeks of Blue capacity, 30 (10x3) weeks of Red capacity and 12 (4x3) weeks of Golden capacity. As stated before, resources are sometimes idle in this mode.
- Resources can be more fully utilized by spreading them across three work-streams. In the “Local Optimization” mode, the total capacity required from every resource is less than in the “Global Optimization” mode. Blue resources can finish all three tasks in 15 weeks instead of 18; Red resources can finish all three tasks in 25 weeks instead of 30; and Golden resources can finish all three tasks in 10 weeks instead of 12.
Let’s now analyze which mode yields better performance.
- CYCLE TIME: In the “Local Optimization” mode, each of the three works-streams takes 50 weeks to get done (15 weeks for the Blue Task + 25 weeks for the Red Task + 10 weeks for the Golden Task). In the “Global Optimization” mode, the cycle time to finish one work stream is 20 weeks.
- THROUGHPUT: In the “Local Optimization” mode, the three work-streams get done in 50 weeks. In the “Global Optimization” mode, the three work-streams finish in 40 weeks (the first work-stream starts immediately and finishes in Week 20, the second work-stream starts after 10 weeks and finishes in Week 30, and the third work-stream starts after 20 weeks and finishes in Week 40). In the “Global Optmization” mode, a fourth work-stream can be started after Week 30 and can be completed by Week 50, a 33% higher throughput.
- MANAGEABILITY: In the “Local Optimization” mode, the number of active work-streams that have to be managed at a time is three. In the “Global Optimization” mode the number of active work-streams that have to be managed is two. Simplification for departmental managers is even more. The number of active tasks that have to be managed at a time comes down from three to one in every department.
“Local Optimization” is the only option for managers when they cannot analyze the impact of local schedules on global performance in real-time. However, “Local Optimization” doesn’t make sense in the new world of “Real-Time Scheduling”.