by F. Robert Jacobs
This study is a Learning Curve[i] analysis of Tesla Model 3 production using data provided by the company[ii]. Using Model 3 production data from 2017-Q3 through 2018-Q3, the Wright learning curve function was fitted. The squared error (difference) between the predicted cumulative days to achieve cumulative production amounts and the reported cumulative production amounts was the criteria used in the fitting.
The two parameters fitted are the learning curve rate, and time for the first unit produced. The analysis indicates that the learning curve rate that Tesla has experienced is approximately 60 percent and that the time for the first unit was approximately 14 days.
The following graph shows the actual versus the predicted time/unit.
As can be seen the fit looks good. The F statistic comparing these two sets of values is only 1.18 which indicates they are not significantly different.
Using the model to predict future Model 3 production indicates that Tesla may be able to work off the order backlog by the end of 2019-Q2.
The most recent estimate of the Model 3 order backlog was 420,000 units. We see that Tesla cumulative production could well exceed this amount by the end of 2019-Q2. If one assumes that the Model 3 is being made on two assembly lines that run 16 hours a day, 5 days a week and 13 weeks/quarter, then the cycle time would be down to approximately 2 minutes per unit at the end of 2019-Q2. This would appear to be feasible. At this point, market demand would dictate production rates for the future and the learning curve rates would not be applicable.
[i] Wright, T. P., “Factors Affecting the Cost of Airplanes,” Journal of Aeronautical Sciences, 3(4)(1936): 122-128.
[ii] Tesla has provided this data in press releases on Oct 2, 2018, Jul 2, 2018, Apr 3, 2018, Jan 3, 2018, etc. The financial data are from Quarterly “update letters from Tesla. The data is available from http://ir.tesla.com/press-releases.