The contribution of learning (Boarin et al., 2012) applies at various levels: a better work organization on the same site, where the personnel have already had experience in the construction and assembling of previous NPP modules; a learning component in factory fabrication of the equipment; a learning component in the utilization of materials and equipment by more skilled workers, etc. A scale-up of the plant output and the attempt to introduce an original French design, i. e. the N4 reactors, towards the end of the program may only partially explain such an occurrence.

Lovins (1986) presented an interesting theoretical framework, referred to as the Bupp-Derian-Komanoff-Taylor hypothesis, that suggests that with increasing application (‘doing’), the complexity of the technology inevitably increases, leading to inherent cost escalation trends that limit or reverse ‘learning’ (cost reduction) possibilities. In other words, the technology scale-up can lead to an inevitable increase in systems complexity that translates into real-cost escalation, or ‘negative learning’. Nevertheless, learning effects have been recorded in technology-advanced industries (Frischtak, 1994); learning effect description was first published by an aeronautical engineer (Wright, 1936).

The learning effect is also visible in the Korean NPP fleet deployment costs: learning accumulation has played an undeniable effect on a progressive cost decrease (Figure 10.10). KHNP, the owner of all 21 of South Korea’s operating nuclear power reactors, has held a licensee relationship with Westinghouse since the late 1980s when the US-based company supplied the 945 MWe System 80 nuclear steam supply for Yonggwang 3 and 4. After that, KHNP was able to develop variants of System 80 for its own requirements under technology transfer terms in the license agreement. After introducing domestic innovations and updating technology over time, KHNP came up with the Korean Standard Nuclear Plant (KNSP), then the OPR-1000. The current APR-1400 technology represents a further evolution of that design. The construction and power generation costs of the APR-1400 are reported to be 10% lower than those of OPR-1000 units.

Korean NPP is the evidence that learning economy may apply to construction costs: in this case, learning effect was achieved through a concentrated construction (Figure 10.11), with the deployment of twin/multiple units on the same site and by avoiding substantial design modification in order to attain PWR plant standardization and control design complexity. (Wolsong NPP are PHWR-CANDU, provided by AECL, as the only exception to the PWR design.)

It may be argued that in principle, learning accumulation is expected to determine a construction cost and time-progressive decrease of successive NPP units, as it was in the Korean NPP fleet. Nevertheless, as far as western countries are considered, in the real world there is often no evidence of cost and time benefits in large NPP deployment programs. That is why simpler and smaller NPPs, with design modularity and high content of factory fabrication, have a higher chance of controlling complexity and exploiting standardization, enabling learning accumulation on both construction and assembling phases. SMRs are expected to benefit from anticipated learning effects, mostly arising from the construction and assembling of multiple units on the same

site. Given the power size of a nuclear site, more SMR units should be fabricated and installed than LRs, with improved chances to learn. General learning accumulation may be recorded at the engineering procurement and construction (EPC) level residing in the human resources knowledge and approach to the project management, and to the organization and procurement issues, such as supplier selection. This learning applies independently of the site location of the new NPP and is therefore indicated as ‘worldwide’ learning in Figure 10.12. In addition, site-level learning accumulation is also applicable on successive NPP units built on the same site, residing in the best, refined practices and actions by local staff. The magnitude of the two effects is comparable (Boarin and Ricotti, 2011a). The learning effect is destined to fade out over the first five to seven units (Carelli et al., 2010). For this reason, in a mature phase of the market, worldwide learning is not a differential factor for SMRs and LRs, while SMRs keep the benefit from the on-site learning accumulation, which applies in case of multiple units built on the same site.