Interactions between wind and solar within the uncertain technology - PowerPoint PPT Presentation

Presented at the 36th USAEE/IAEE Conference Interactions between wind and solar within the uncertain technology ecological system DUAN Hongbo University of Chinese Academy of Sciences, China University of Kansas, USA Washington, D . C, Sep. 25, 2018

CONTENTS 1 Motivations 2 Proposed model 3 Data and estimation 4 Main results and analysis 5 Concluding remarks

Background • The carbon budget under the 2°C target will run out in the coming twenty years, we actually have few choices but resorting to energy transitions from carbon-based energy to renewables. • As the latest IRENA report states, wind and solar may dominate the primary energy demand during the achievement of this goal. • Wind power and PV solar energy markets skyrocket in recent years, and the global cumulative installed capacity of PV solar has expanded from 6 GW in 2006 to 303 GW in 2016, annually growing by 148%,versus 121% for wind power. • Many countries lead this trends, especially for China and India.

Cumulative capacity New added capacity Cumulative capacity New added capacity 350 80 600 70 China China 70 300 60 US 500 US 60 250 Japan Germany 50 400 Gigawatts � Gigawatts � Gigawatts � Gigawatts � 50 India India 200 40 ROTT 40 ROTT 300 150 30 ROW ROW 30 200 100 20 20 100 50 10 10 0 0 0 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Fig. 1. Trends of global PV solar and wind power development

Motivations • Diffusion and competitive substitution of key renewable energy technologies substantially affect the dynamic evolution of energy structure, which in turn poses great effects on mitigating greenhouse gas (GHG) emissions and tackling climate change. However the extant research is limited: — Analysis on diffusion of product innovation in social system has long been the center of public and academic interest , very few attentions has been paid on energy technology, particularly for possible interactions — It’s of great value to relax the deterministic nature of the dynamic system models to inform the potential impacts of random disturbances on technology diffusion and external interactions

Missions • First, based on population ecology theory, we dedicate to develop a stochastic analytic framework of technology diffusion and interaction, i.e., the randomized Lotka-Volterra model by taking the impacts of disturbances into account. • The second mission is to make a cross-country analysis on wind power and PV solar’s long-term technology penetration patterns, as well as the possible dynamic interactions between these two technologies under random perturbations. • Third, short-term forecasts are also provided based on the proposed stochastic dynamic model framework.

Model framework ➣ The typical L-V model ² The proposed stochastic L-V model Given initial value X(0)=(x 1 (0), x 2 (0)) T ( x 1 (0), x 2 (0)>0 ), X(t)=(x 1 (t), x 2 (t)) T is the solution of Model (2) ; σ i 2 (i=1,2) is white noise disturbance, B i (t) (i=1,2) are independent geometric Brownian motions.

Mode - derived interactions a 21 �� a 12 � ( a 21 � a 12 ) �

Data and estimation method 200 q Country set: China, the US, Wind (GW) � Japan, the UK, India, 150 France, Italy, Germany, 100 Spain, Canada, Sweden, the 50 Netherlands; 0 q Data sources: Wind installed 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 capacity is mainly from 140 120 GWEC, while PV data are Solar (GW) � UK US China India France Italy Germany Spain 100 abstracted from IEA, BP and 80 PV Magazine ; 60 40 q We adopt MLE method to 20 0 estimate the stochastic L- 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 V model on Matlab 2016 software platform. UK US China India France Italy Germany Spain Fig. 3. Growth trends of wind and PV solar markets for the top 8 countries

Main results: model estimation

Main results: model fitting

Main results: self-interaction Table 2. Identification of equilibrium points and stability under Model (1) across countries

Main results: external interactions q T he deficit in China has accumulated to more than 100 billion RMB by 2017 , which yields remarkable adverse effects on wind technology, and that’s why 20%-30% of wind power has been abandoned in the wind areas. q The market positions of PV solar and wind power technology in the US and a 21 � a 12 � Sweden keep the same with those in China, i.e., PV solar perform as predator, versus prey for wind power technology. ; q And the policy environment provide the possible explanations.

Main results: equilibrium analysis (1) Table 1. Technology characteristics for both wind and PV solar across countries

Main results: equilibrium analysis (2)

Main results: short-term forecasts 1.2 15 2015-2017 � 2018-2020 � 2015-2017 � 2018-2020 � Percentage (%) Percentage (%) 1 0.8 10 0.6 0.4 5 0.2 0 0 India Japan Italy India Japan Italy India Japan Italy India Japan Italy Fig. 5. Short-term forecasts for stably vibrated countries

Main results: prediction accuracy q India: Wind trends will Table 4 . Summary for prediction accuracy across countries be reinforced (113.9%); Solar growth decays , still as high as 142.9%. q Japan: Wind keeps the pave of the past 3 years (95.8%), versus 40.4% annual increase of solar PV market. q Italy: the markets with the lowest annual growth (11.5% and 8.3%), further reduce to 1.3% and 1.8%.

Concluding remarks ² We find both positive and negative scale effects for wind markets, while PV solar markets are consistently scale- restrictive for all the target countries . ² The current technology interactions are dominated by mutualism and prey-predator types, of which prey-predator relationships mainly exist in the US, China and Italy, with PV solar to be predator. ² Random technology orbits for both wind and PV solar oscillate around the deterministic equilibrium orbit, normally distributed, and the mean orbit of such large-scale random orbits converges to the analytic equilibrium orbit.

Q & A Assoc. Prof. Hongbo Duan hbduan@ucas.ac.cn hb.duan@ku.edu