Comparative Evaluation of Learning Curve Models for Construction Productivity Analysis

Welder learning curves behaviour: “focus on welding productivity with the TIG process of marine platforms stainless steel pipes”

“…Additionally, the study of the behaviour of the curves elaborated based on AP data did not detect an improvement in global productivity. This result is in agreement with the literature, both for welding and other activities in the construction industry (Ammar and Samy, 2015;Anzanello, 2004;Cadaval J unior et al, 2013;Pellegrino et al, 2012;Jarkas, 2016;Pellegrino and Constantino, 2018;Raman and Varghese, 2016;Srour et al, 2016;Ralli et al, 2020). According to Martins (2011) and PROMINP (2009), improvement in productivity did not occur after the tenth day.…”

“…The recursive model and graph based on unitary data are the most appropriate for short-term projects. Ralli et al (2020) realised a comparative study among models such as potential, Stanford "B", cubic; piecewise or stepwise, and exponential, during the manufacturing of floating marine concrete caissons in large projects. In this study, they considered a graphic representation with unitary and cumulative data.…”

“…According to Martins (2011) and PROMINP (2009), improvement in productivity did not occur after the tenth day. It was expected that this improvement would occur owing to the acclimatisation of workers to the conditions of the workplace (Pellegrino and Constantino, 2018;Ralli et al, 2020). Notably, pipe welding is an individual's activity, depending on the skill of each welder, and the sample studied considers the daily average of the welders and not the product of teamwork.…”

“…Conversely, despite the exponential and potential models presenting slightly close adjustment conditions, the performance of the former was slightly superior. Thus, owing to the popularity of the potential model, it could be adopted for obtaining acceptable results (Ralli et al, 2020). As for the sample composition, the results considering the curves elaborated from the AP for the potential and exponential models resulted in reasonable quality adjustments for the diameter ranges from 1" to 2 7/8'' and higher than 2 7/8''.…”

Factors that Influence the Learning Curve: Evidence from Cost Behavior in Clinical Labs*

“…4 4 The cumulative average learning curve model is of the form $$$ AC=a{x}^{-b} $$$, where $$$ a $$$ is the hours or cost of producing the first unit, $$$ AC $$$ is the cumulative average hours or cost of producing $$$ x $$$ units, and $$$ b $$$ measures the learning rate. This model is extensively used in archival, experimental, and simulation research (Grosse et al 2015; Glock et al 2019), and by regulatory agencies like the Department of Defense (DoD) and National Aeronautics and Space Administration (NASA) to forecast costs, which are used in budgeting and performance evaluation (Moore et al 2015; Brown and Anderson 2019; DoD 2020; Ralli et al 2020). …”