From single-factor to multi-factor learning curves for modelling innovation -a review
1 like777 views
This document discusses modeling energy technology cost reductions using learning curves. It begins by introducing single-factor learning curves that relate cost reductions to deployment levels. However, these cannot explain differences between emerging and current technologies or various drivers of global versus local learning. The document then reviews literature using multi-factor learning curves that incorporate additional drivers such as R&D, supply chain interactions, economies of scale, and component-level costs. These provide a more comprehensive understanding of innovation challenges but require more data and have limitations from model complexity and variable correlation. In conclusion, multi-factor learning curves are an improved approach for analyzing technology cost reductions.
From single-factor to multi-factor learning curves for modelling innovation -a review
- 1. Environmental Research Institute University College Cork From single-factor to multi-factor learning curves for modelling innovation - a review Alessia Elia, Fionn Rogan, Brian O Gallachoir ETSAP Workshop, Stuttgart, GERMANY | Nov 9th 2018
- 2. What is happening energy technology costs (e.g. PV)?
- 3. How to model cost reductions due to energy technology learning? Basic method: SINGLE FACTOR LEARNING CURVE 𝐶𝑡 = 𝑓 𝑄 = 𝑎 𝑄β 𝐶𝑡 is unit costs 𝑄 is learning by-doing factor (e.g. capacity installation) β is elasticity of learning learning rate 𝐿𝑅 = 𝐹 β = 1 − 2β Relates cost reductions of a technology to deployment
- 4. What drove cost reductions of PV from 1979 to 2001? Source: (Nemet, 2006)
- 5. How many academic papers use multi-factor learning curves? Energy Technology Cost Reduction Drivers Number of Papers Learning By Doing 33 Learning By Researching 19 Learning by Interacting (geographical) 6 Learning by Interacting - Industry 6 Input material 13 Economies of scale 16 External learning - Policy 7 External learning - Market 5 External learning - Resource/technology 9 114
- 6. What learning dynamics are not explained in 1FLCs? 1. Differences between Emerging Technologies vs Current Technologies 2. Drivers impacting Global Learning and Local Learning 3. Drivers impacting Market-Wide vs Industry-Wide learning 4. Different costs components learning • Technology Vs one-component learning vs system learning
- 7. How does learning differ for emerging vs current technologies? Other drivers ECONOMIES OF SCALE POLICY SUPPORT CLUSTER FORMATION & SPILLOVERS
- 8. What is the difference between global vs local learning? Location 1 Loc. 1 Location 2 Loc. 3 Location 3 Loc. 2 Global WIND ONSHORE 1) Research in different countries during 1970s (DENMARK, US, GERMANY) 2) Collaboration and deployment in niche market during 1980s (e.g. Danish turbines deployed in US) Local deployment achievements Capacity deployment in-country 3) Delocalization manufactures, market expansion in new interesting markets 1 2 3 New market s KNOWLEDGE SPILLOVER IN NEW COUNTRIES Research knowledge, technical achievement Manufacturing learning INTERACTION BETWEEN COUNTRIES and ACTORS Technical achievement Manufactures learning
- 9. What is difference between Industry-wide Vs Market-wide learning and interaction? Public R&D Policy Manufacturing learning Company 1 Market 1 Company n Industry-wide learning Company 2 industry R&D Regulation Market changes Energy quality Market n Feedback from the market Feedback Market-wide learning R&D LEARNING and SUPPORT MANUFACTURING DEPLOYMENT EXPERIENCE CLUSTER & SPILLOVERS INNOVATION: MATERIAL USE AND TECHNOLOGY SCALE POLITICAL and NATIONAL STABILITY DEVELOPMENT OF MARKET STRUCTURE e.g. infrastructures Fees, exchange rate, inflation, input material price changes MARKET SPILLOVERS and NETWORKING Source: Adapted from Martinsen 2009, based on figure 2.2 and 2.9 in IEA (2000), figure 2.3 and 3.1 in Neij et al. (2003).
- 10. How does learning vary by costs components? technology vs multi-component learning vs system learning PV module panel Balance of the system: Inverter Mounting system Electrical cabling Project plant costs + Operating costs Silicon, Silver price Energy consumptions Capital costs PV manufactures Labour costs Technology costs Hard/Soft deployment costs: Installation, transportation, project design, customer acquisition Financing, support scheme Input costs Technology costs Maintenance work (cleaning, inspection, repair) Insurance Metering Legal-administrative work (e.g. processes related to taxation) Decommissioning Energy prices Previous costs Energy performance (Capacity factor, technology life) Financial return Single Factor Learning Curves cannot explain all cost reductions Risk of overestimation of deployment effect and underestimation of other drivers
- 11. Multi-factor learning curves advantages Includes broader set of factors impacting costs reduction: • Value of innovation activities and R&D investments • Value of spillover of knowledge between countries • Value of interaction between different actors, knowledge exchange between partners in same industry • Effect of economies of scale of a device, plant, or an industry • Material use change, market price changes
- 12. Multi-factor learning curves summary • Multi factor learning curve models can incorporate more information to understand the innovation challenges • 2-factor experience curves shed new light on value of R&D investment and innovation activity on renewable energy • Learning by-interacting driver explore the effect of industry formation on costs reduction • Drawback: limits on data access, model limitation as dangers of over fitting, correlation between learning factors, causation between dependent and independent variables
- 13. Environmental Research Institute University College Cork Thank you Alessia Elia (alessia.elia@ucc.ie) Fionn Rogan (f.rogan@ucc.ie) Brian O Gallachoir (b.ogallachoir@ucc.ie)