- Problem Definition
In Garuda AACE 2015 Project, on Week 10, the Author and 4 other members, submitted Change Order (CO) to Client for descoping Paper Project. The Client accepted the CO with notes that those members need to do rebaseline on W11. Before rebaseline, the Client requested us to apply Learning Curve on the project , in order to have sufficient input for rebaseline. In this week’s blog, the Author will apply Learning Curve on two running project : Weekly Report and Blog.
- Development of Feasible Alternatives
The budget manhours plan and actual recorded (cut of week 11) of Author’s Weekly Report as per below :
Table 1. Manhours Plan vs Actual of Weekly Report (as is)
The budget manhours plan and actual recorded of Author’s Blog Project as per below :
Table 2. Manhours Plan vs Actual of Blog Project (as is)
To gain more accurate data, some outlier data (yellow highlight) are excluded from the data source. Table 1 outliers are come from extra works to revise the Team Report. Table 2 outliers are come from extra learning of first blog and zero deliverable or no blog submission on week 8 and 11. The data now as per table below :
Table 3. Manhours Plan vs Actual of Weekly Report (clean)
Table 4. Manhours Plan vs Actual of Blog Project (clean)
- Development of the Outcomes for Alternativea
The concept of a Learning Curve is each time the cumulative production doubles, the hours required to produce the most recent unit decreases by approximately the same percentage, known as a slope.
Refer to the power trend line approach (Y=ax-b) as explained in Sullivan’s Engineering Economy:
Zu = K(un) 
u = the output unit number
Zu = the number of input resource units needed to produce output unit u
K = the number of input resource units needed to produce the first output unit
S = the learning curve slope parameter expressed as a decimal (can be calculated S = 2n)
n = log s/log 2 = learning curve exponent
Figure 1: Power Trendline approach on Weekly Report
Figure 2: Power Trendline approach on Blog
- Selection of the Acceptable Criteria
The purpose of this learning curve analysis is to estimate realistic manhours based on previous actual hours and current learning curve parameter that required to complete the task. Basic expectation is realistic efficiency will required less manhours to produce the assigned tasks. However the less efficiency will also be consider for more manhours if realistic actual data stated that condition.
- Analysis and Comparison of the Alternatives
Figure 3: Plot Unit Time and Cumulative Average on Weekly Report
Figure 4: Plot Unit Time and Cumulative Average on Blog
- Selection of the Preferred Alternative
Based on the Table 5 above, with the current level of efficiency, total hours is estimated decrease to 21.56 hours with average manhours per report is 0.94 or rounded to 1 hours per report.
And based on Table 6 above, with the current level of efficiency, total hours is estimated increase to 63.57 hours with average manhours per report is 2.69 or rounded to 2.7 hours per blog.
- Performance Monitoring and Post-Evaluation of Results
Periodically along the actual data grows, similar analysis could be perform (including analysis to the other task). We can also perform the Estimate at Completion (EAC) approach, to check the result of Learning Curve Analysis above, in order to provide more confidence on the estimate. The application of EAC for above cases will be done in next blog
- Sullivan, W.G., Wick, E. M., Koelling, C.P. (2012). Engineering Economy, Fifteenth Edition, Pearson, Chapter 3, 108-111.
- Humphreys, G. C. (2002). Lerning Curve. Project Management Using Earned Value (3rd ed.). Orange, CA: Humphreys & Assoc.
- Darwito, R. (2015) W09_RD_ Team Weekly Report Learning Curve . Retrieved from : https://garudaaace2015.wordpress.com/2015/05/03/w09_rd_-team-weekly-report-learning-curve/