Small is Better: The Benefits of Granularity in Energy Technologies

Transcription

1 Small is Better: The Benefits of Granularity in Energy Technologies Charlie Wilson (with Arnulf Grubler, Nuno Bento, Caroline Zimm...) January 2018 Science Policy Research Unit (SPRU), University of Sussex

2 Small is... has a long tradition in technology studies, particularly for distributed energy-supply systems

3 granular small unit size low unit cost modular replication lumpy large unit size high unit cost indivisible up-scaling

4 Unit size and unit cost strongly correlate in diverse samples of energy supply and end-use technologies 1.E+10 Granularity metrics: unit scale vs investment cost upscaling vs. modular 1.E+10 Granularity metrics: unit scale vs investment cost end-use vs. supply Investment cost per unit ($2009) 1.E+09 1.E+08 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 upscaling modular Investment cost per unit ($2009) 1.E+09 1.E+08 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 energy supply energy end-use Power (energy supply) Power (energy end-use) y = x R² = y = 2937x R² = E+02 1.E+02 1.E+01 1.E+01 1.E+00 1.E+00 1.E-03 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 1.E-03 1.E-01 1.E+01 1.E+03 1.E+05 1.E+07 granular Average unit size (kw) lumpy granular Average unit size (kw) lumpy end-use tends to be more granular supply tends to be more lumpy

5 Are granular energy technologies better? lower adoption risks? more rapid learning rates? faster diffusion times? lower risks of lock-in? more equitably distributed? innovation and diffusion processes system outcomes

6 Granularity (1): lower adoption effort (investment per unit) results in faster diffusion ( t) diffusion of 35 industrial, energy, transport, and consumer good innovations (US) 35% of variance in t explained by investment size granular lumpy NB. two outliers exclude: cars + 2 * WW2

7 Size matters. Megaprojects carry large risks associated with complexity, one-off designs, and long lead times The iron law of megaprojects [Flyvberg 2014]: they run over budget, over time, over and over again Adoption risks with lumpy technologies: (i) bespoke (non-standard) design limits learning; (ii) complexity, interdependencies, interoperability challenges; (iii) long planning horizons create exposure to exogenous change; (iv) involvement of diverse actors with competing interests. "policymakers should prefer energy alternatives that require less upfront outlays and that can be built very quickly [Ansar et al. 2013].

8 Granularity (2): smaller unit sizes & modularity result in lower adoption risk (% cost overrun) size positively correlated with cost overruns in 7 of 9 samples for up-scaling technologies nuclear thermal hydro size negatively correlated with cost overruns in 4 of 5 samples for modular technologies wind solar granular lumpy Reanalysis of data from: Sovacool et al. (2014). Energy Research & Social Science 3:

9 x 1 x 1,000 Some technologies are more open to improvement than others. Compact, modular systems, such as photovoltaics and electronics, are easily experimented on... Trancik (2014). Nature 507: smaller units -> more units -> more opportunities to experiment & learn x 1,000,000 -> higher rates of cost reduction -> more units

10 Higher learning rates (on average) are associated with standardised production of large number of units Learning(Rate((%)( 40$ 30$ 20$ 10$ 0$!10$!20$!30$ Learning(Rates(vs.(Cumula2ve(Produc2on( small n units mean LR 10% nuclear (excluded) large n units mean LR 20%!40$ 1E+02$ 1E+03$ 1E+04$ 1E+05$ 1E+06$ 1E+07$ 1E+08$ 1E+09$ 1E+10$ 1E+11$ 1E+12$ Cumula2ve(Number(of(Units(Produced((nearest(Order(of(Magnitude)( Transistors$(World)$ DRAMs$(World)$ Automobiles$(World)$ Washing$machines$(World)$ Refrigerators$(World)$ Dishwashers$(World)$ Freezers$(upright)$(World)$ Freezers$(chest)$(World)$ Hand!held$calculators$(US)$ Compact$fluorescent$light$bulbs$(US)$ Dryers$(World)$ Air$condiRoning$&$heat$pumps$(US)$ Air$furnaces$(US)$ Solar$hot$water$heaters$(US)$ PV$modules$(World)$ Wind$turbines$(World)$ Heat$pumps$(Sweden,$Switzerland)$ Gas$turbines$(World)$ Pulverized$coal$boilers$(World)$ Hypropower$plants$(OECD)$ Nuclear$reactors$(US,$France)$ Ethanol$(Brazil)$ Coal$power$plants$(OECD)$ Coal$power$plants$(US)$ Gas$pipelines$(US)$ Gas$combined$cycles$(OECD)$ Hydrogen$producRon$(SMR)$(World)$ LNG$producRon$(World)$

11 Granularity (3): more unit numbers enable higher learning rates (controlling for unit scale economies) De-scaled Learning Rate (CumulaBve Number of Units) 25% 20% 15% 10% 5% 0% -5% -10% -15% -20% Learning rates per doubling of cumula<ve # of units controlling for unit economies of scale (exc. 2 outliers) y = ln(x) R² = % 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 granular Average Unit Size (MW) lumpy unit scale is a stronger predictor of learning rate after controlling for economies of scale NB. two outliers excluded: -35% nuclear +32% geothermal Data from: Healey, S. (2015). Separating Economies of Scale and Learning Effects in Technology Cost Improvements. IR International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria.

12 Granularity (7): shorter lifetimes of smaller units enable rapid turnover and reduce risk of lock-in Average technical lifecme (years) y = 1.777ln(x) R² = E E E E E E+07 granular Granularity (unit size) vs. technical lifecme Average unit size at to (kw) lumpy lock-in = resistance to change in technological systems causes: - technological - institutional - behavioural granularity: - shorter lifetimes - more rapid innovation cycles

13 Granularity (7): lower complexity (interdependencies) of smaller units further reduce risk of lock-in lock-in = resistance to change in technological systems causes: - technological - institutional - behavioural granularity: - lower complexity, (as measure of interdependency) granular lumpy Component data from: Ayres (1988). Manufacturing Review 1(1):

14 Lorenz curves can describe distribution of access to useful technologies (and service infrastructures) Lorenz curves: distribution of access to technologies NB1. includes non-access NB2. not all countries: ~6bn people Data from: Zimm, C. (2017). International Institute for Applied Systems Analysis (IIASA), Laxenburg, Austria.

15 Granularity (8): lower barriers to adoption result in more equitably distributed access to useful services Gini coefficient = measure of distributional (in)equality calculated from Lorenz curves 0 = perfect equality 1 = perfect inequality more granular = lower cost per additional access

16 In sum: Granularity has many generalizable benefits technology characteristics adoption environments lower adoption effort & faster diffusion times [1] lower adoption risks [2] shorter formative phases [4] more rapid learning rates [3] faster spatial diffusion [6] larger market sizes [5] lower risks of lock-in [7] greater benefits for system efficiency [9] more equitable distribution [8] higher social legitimacy [10] innovation and diffusion processes system outcomes

17 But benefits of granularity depend on replication, standardisation...

18 But benefits of granularity depend on replication, standardisation... and access to infrastructure granularity benefits production, manufacturing (standardisation, serialisation) installation, adoption (learning, accessibility) required conditions dominant designs homogeneous producers repetitive installation low skill adoption distributed, modular infrastructure potential issues experimentation & variety heterogeneous producers bespoke installation high skill adoption system-wide, lumpy infrastructure other more general issues with granularity: (1) transaction costs; (2) dispersed impacts; (3) lifecycle impacts...

19 Granularity is not a hegemonic strategy... but it is too often a marginalised one Times editorial 1 December 1977 Dr Schumacher did not advocate smallness as the answer to everything. The title of his book has misled many people. What he was talking about was the appropriate size for different structures some large, some small. He concentrated only on smallness only to counteract the idolatry of gigantism.

20 Small is Better: The Benefits of Granularity in Energy Technologies Charlie Wilson (with Arnulf Grubler, Nuno Bento, Caroline Zimm...) January 2018 Science Policy Research Unit (SPRU), University of Sussex