From 28e541dcd85fad014774e26bafc684f982ba0527 Mon Sep 17 00:00:00 2001 From: Jihoon Min Date: Sat, 24 Aug 2019 12:00:30 +0200 Subject: [PATCH 1/3] Suggested changes for sections: energy demand & tech change Minor edits and improvements In tech change section, it would be helpful to add remarks on how in general diffusion constraints are determined for different technologies or what the values are based on. In energy demand section, frequent uses of 'several' sound too ambiguous. --- source/energy/demand.rst | 18 ++++++++++++++---- source/energy/tech.rst | 8 ++++---- 2 files changed, 18 insertions(+), 8 deletions(-) diff --git a/source/energy/demand.rst b/source/energy/demand.rst index 479db02..bd47df5 100755 --- a/source/energy/demand.rst +++ b/source/energy/demand.rst @@ -13,8 +13,8 @@ service demands that are provided to MESSAGE, including: 6. Transportation 7. Non-commercial biomass. -These demands are generated using a so-called scenario generator which is implemented in the script language `R `_. The scenario generator uses country-level -historical data of GDP per capita (PPP) and final energy use as well as projections of GDP (PPP) and population to extrapolate the seven energy service demands into the future. The +These demands are generated using a so-called scenario generator which is implemented in the script language `R `_. The scenario generator relates historical country-level +GDP per capita (PPP) to final energy and, using projections of GDP (PPP) and population, extrapolate the seven energy service demands into the future. The sources for the historical and projected datasets are the following: 1. Historical GDP (PPP) – World Bank (World Development Indicators 2012 :cite:`world_bank_group_world_2012`) @@ -23,9 +23,19 @@ sources for the historical and projected datasets are the following: 4. Projected GDP (PPP) – Dellink et al (2015 :cite:`dellink_long-term_2015`), see Shared Socio-Economic Pathways database (`SSP scenarios `_) 5. Projected Population – KC and Lutz (2014 :cite:`kc_human_2014`), see Shared Socio-Economic Pathways database(`SSP scenarios `_) -The scenario generator runs regressions on the historical datasets to establish the relationship between the independent variable (GDP (PPP) per capita) and several dependent variables, including total final energy intensity (MJ/2005USD), the shares of final energy in several energy end-use sectors (transport, residential/commercial and industry) and the shares of electricity use in the industrial and residential/commercial sectors. In the case of final energy intensity, the relationship is best modeled by a power function so both variables are log-transformed. In the case of most sectoral shares, only the independent variable is log-transformed. The exception is the industrial share of final energy, which uses a hump-shaped function inspired by Schafer (2005) :cite:`schafer_structural_2005`. This portion of the model provides the historical relationships between GDP per capita and the dependent variables for each of the eleven MESSAGE regions. +The scenario generator runs regressions on the historical datasets to establish the relationship for each of the eleven MESSAGE regions between the independent variable (GDP (PPP) per capita) and multiple dependent variables: -The historical data are used in `quantile regressions `_ to develop global trend lines that represent each percentile of the cumulative distribution function (CDF) of each regressed variable. Given the regional regressions and global trend lines, final energy intensity and sectoral shares can be extrapolated based on projected GDP per capita, or average income. Several user-defined inputs allow the user to tailor the extrapolations to individual socio-economic scenarios. In the case of final energy intensity (FEI), the extrapolation is produced for each region by defining the quantile at which FEI converges (e.g., the 20th percentile) and the income at which the convergence occurs. For example, while final energy intensity converges quickly to the lowest quantile (0.001) in SSP1, it converges more slowly to a larger quantile (0.5 to 0.7 depending on the region) in SSP3. Convergence quantiles and incomes are provided for each SSP and region in :numref:`tab-quantssp1`, :numref:`tab-quantssp2`, :numref:`tab-quantssp3`. The convergence quantile allows one to identify the magnitude of FEI while the convergence income establishes the rate at which the quantile is approached. For the sectoral shares, the user can specify the global quantile at which the extrapolation should converge, the income at which the extrapolation diverges from the regional regression line and turns parallel to the specified convergence quantile (i.e., how long the sectoral share follows the historical trajectory), and the income at which the extrapolation converges to the quantile. Given these input parameters, the user can extrapolate both FEI and sectoral shares. +1. Total final energy intensity (MJ/2005USD) +2. Shares of final energy among several energy end-use sectors (transport, residential/commercial and industry) +3. Shares of electricity use between the industrial and residential/commercial sectors. + +In the case of final energy intensity, the relationship is best modeled by a power function so both variables are log-transformed. In the case of most sectoral shares, only the independent variable is log-transformed. +The exception is the industrial share of final energy, which uses a hump-shaped function inspired by Schafer (2005) :cite:`schafer_structural_2005`. + +In parallel, the same historical data are used, now globally, in `quantile regressions `_ to develop global trend lines that represent each percentile of the cumulative distribution function (CDF) of each dependent variable. Given the regional regressions and global trend lines, final energy intensity and sectoral shares can be extrapolated based on projected GDP per capita, or average income. + +A basic assumption here is that the regional trends derived above will converge to certain quantiles of the global trend when each region reaches a certain income level. Several user-defined inputs allow users to tailor the extrapolations to individual socio-economic scenarios. +In the case of final energy intensity (FEI), the extrapolation is produced for each region by defining the quantile at which FEI converges (e.g., the 20th percentile within the global trend) and the income at which the convergence occurs. For example, while final energy intensity converges quickly to the lowest quantile (0.001) in SSP1, it converges more slowly to a larger quantile (0.5 to 0.7 depending on the region) in SSP3. Convergence quantiles and incomes are provided for each SSP and region in :numref:`tab-quantssp1`, :numref:`tab-quantssp2`, :numref:`tab-quantssp3`. The convergence quantile allows one to identify the magnitude of FEI while the convergence income establishes the rate at which the quantile is approached. For the sectoral shares, users can specify the global quantile at which the extrapolation should converge, the income at which the extrapolation diverges from the regional regression line and turns parallel to the specified convergence quantile (i.e., how long the sectoral share follows the historical trajectory), and the income at which the extrapolation converges to the quantile. Given these input parameters, users can extrapolate both FEI and sectoral shares. The total final energy in each region is then calculated by multiplying the extrapolated final energy intensity by the projected GDP (PPP) in each time period. Next, the extrapolated shares are multiplied by the total final energy to identify final energy demand for each of the seven energy service demands used in MESSAGE. Finally, final energy is converted to useful energy in each region by using the average final-to-useful energy efficiencies used in the MESSAGE model for each model region (:ref:`spatial`). diff --git a/source/energy/tech.rst b/source/energy/tech.rst index 0ed2529..4f4fd1e 100755 --- a/source/energy/tech.rst +++ b/source/energy/tech.rst @@ -2,7 +2,7 @@ Technological change ====================== -Technological change in MESSAGE is generally treated exogenously, although pioneering work on the endogenization of technological change via learning curves in energy-engineering type models (Messner, 1997 :cite:`messner_endogenized_1997`) and the dependence of technology costs on market structure has been done with MESSAGE (Leibowicz, 2015 :cite:`leibowicz_growth_2015`). The current cost and performance parameters, including conversion efficiencies and emission coefficients are generally derived from the relevant engineering literature. For the future, alternative cost and performance projections are developed to cover a relatively wide range of uncertainties that influence model results to a good extent. +Technological change in MESSAGE is generally treated exogenously, although pioneering works on the endogenization of technological change via learning curves in energy-engineering type models (Messner, 1997 :cite:`messner_endogenized_1997`) and the dependence of technology costs on market structure have been done with MESSAGE (Leibowicz, 2015 :cite:`leibowicz_growth_2015`). The current cost and performance parameters, including conversion efficiencies and emission coefficients are generally derived from the relevant engineering literature. For the future, alternative cost and performance projections are developed to cover a relatively wide range of uncertainties that influence model results to a good extent. Technology cost ---------------- @@ -18,8 +18,8 @@ MESSAGE tracks investments by vintage, an important feature to represent the ine technologies and switch to more suitable alternatives. An important factor in this context that influences technology adoption in MESSAGE are technology diffusion constraints. Technology diffusion in MESSAGE is determined -by dynamic constraints that relate the construction of a technology added or the activity (level of production) of a technology in a period t to construction or the -activity in the previous period t-1 (Messner and Strubegger, 1995 :cite:`messner_users_1995`, cf. section :ref:`upper_dynamic_constraint_capacity`). +by dynamic constraints that relate the construction of a technology added or the activity (level of production) of a technology in a period *t* to construction or the +activity in the previous period *t-1* (Messner and Strubegger, 1995 :cite:`messner_users_1995`, cf. section :ref:`upper_dynamic_constraint_capacity`). While limiting the possibility of flip-flop behavior as is frequently observed in unconstrained Linear Programming (LP) models such as MESSAGE, a drawback of such hard growth constraints is that the relative advantage of some technology over another technology is not taken into account and therefore even for very competitive technologies, @@ -27,7 +27,7 @@ no acceleration of technology diffusion is possible. In response to this limitat (Keppo and Strubegger, 2010 :cite:`keppo_short_2010`). These allow faster technology diffusion at additional costs and therefore generate additional model flexibility while still reducing the flip-flop behavior and sudden penetration of technologies. -:numref:`fig-difconstraint` below illustrates the maximum technology growth starting at a level of 1 in year t=0 for a set of five diffusion constraints which jointly lead to a soft constraint. +:numref:`fig-difconstraint` below illustrates the maximum technology growth starting at a level of 1 in year *t*=0 for a set of five diffusion constraints which jointly lead to a soft constraint. .. _fig-difconstraint: .. figure:: /_static/diffusion_constraint_example.png From 4ac3947843c5c016201f79c3d70f948ebc2b662e Mon Sep 17 00:00:00 2001 From: Jihoon Min Date: Sat, 24 Aug 2019 12:19:15 +0200 Subject: [PATCH 2/3] Additional changes to demand --- source/energy/demand.rst | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/source/energy/demand.rst b/source/energy/demand.rst index bd47df5..324e695 100755 --- a/source/energy/demand.rst +++ b/source/energy/demand.rst @@ -23,7 +23,7 @@ sources for the historical and projected datasets are the following: 4. Projected GDP (PPP) – Dellink et al (2015 :cite:`dellink_long-term_2015`), see Shared Socio-Economic Pathways database (`SSP scenarios `_) 5. Projected Population – KC and Lutz (2014 :cite:`kc_human_2014`), see Shared Socio-Economic Pathways database(`SSP scenarios `_) -The scenario generator runs regressions on the historical datasets to establish the relationship for each of the eleven MESSAGE regions between the independent variable (GDP (PPP) per capita) and multiple dependent variables: +The scenario generator runs regressions on the historical datasets to establish the relationship for each of the eleven MESSAGE regions between the independent variable (GDP (PPP) per capita) and the following dependent variables: 1. Total final energy intensity (MJ/2005USD) 2. Shares of final energy among several energy end-use sectors (transport, residential/commercial and industry) @@ -34,13 +34,13 @@ The exception is the industrial share of final energy, which uses a hump-shaped In parallel, the same historical data are used, now globally, in `quantile regressions `_ to develop global trend lines that represent each percentile of the cumulative distribution function (CDF) of each dependent variable. Given the regional regressions and global trend lines, final energy intensity and sectoral shares can be extrapolated based on projected GDP per capita, or average income. -A basic assumption here is that the regional trends derived above will converge to certain quantiles of the global trend when each region reaches a certain income level. Several user-defined inputs allow users to tailor the extrapolations to individual socio-economic scenarios. +A basic assumption here is that the regional trends derived above will converge to certain quantiles of the global trend when each region reaches a certain income level. Hence, two key user-defined inputs allow users to tailor the extrapolations to individual socio-economic scenarios: convergence quantile and the corresponding income. In the case of final energy intensity (FEI), the extrapolation is produced for each region by defining the quantile at which FEI converges (e.g., the 20th percentile within the global trend) and the income at which the convergence occurs. For example, while final energy intensity converges quickly to the lowest quantile (0.001) in SSP1, it converges more slowly to a larger quantile (0.5 to 0.7 depending on the region) in SSP3. Convergence quantiles and incomes are provided for each SSP and region in :numref:`tab-quantssp1`, :numref:`tab-quantssp2`, :numref:`tab-quantssp3`. The convergence quantile allows one to identify the magnitude of FEI while the convergence income establishes the rate at which the quantile is approached. For the sectoral shares, users can specify the global quantile at which the extrapolation should converge, the income at which the extrapolation diverges from the regional regression line and turns parallel to the specified convergence quantile (i.e., how long the sectoral share follows the historical trajectory), and the income at which the extrapolation converges to the quantile. Given these input parameters, users can extrapolate both FEI and sectoral shares. The total final energy in each region is then calculated by multiplying the extrapolated final energy intensity by the projected GDP (PPP) in each time period. Next, the extrapolated shares are multiplied by the total final energy to identify final energy demand for each of the seven energy service demands used in MESSAGE. Finally, final energy is converted to useful energy in each region by using the average final-to-useful energy efficiencies used in the MESSAGE model for each model region (:ref:`spatial`). .. _tab-quantssp1: -.. table:: Convergence quantile and income for each parameter and region for SSP1 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each variable and region for SSP1 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | @@ -123,7 +123,7 @@ The total final energy in each region is then calculated by multiplying the extr .. _tab-quantssp2: -.. table:: Convergence quantile and income for each parameter and region for SSP2 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each variable and region for SSP2 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | @@ -206,7 +206,7 @@ The total final energy in each region is then calculated by multiplying the extr .. _tab-quantssp3: -.. table:: Convergence quantile and income for each parameter and region for SSP3 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each variable and region for SSP3 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | From 157c0f8c144d9b3096ddea09a8fdffa6e643a754 Mon Sep 17 00:00:00 2001 From: Jihoon Min Date: Wed, 9 Oct 2019 12:01:02 +0200 Subject: [PATCH 3/3] Suggested change from Jarmo --- source/energy/demand.rst | 6 +++--- source/energy/tech.rst | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/source/energy/demand.rst b/source/energy/demand.rst index 324e695..9bdea10 100755 --- a/source/energy/demand.rst +++ b/source/energy/demand.rst @@ -40,7 +40,7 @@ In the case of final energy intensity (FEI), the extrapolation is produced for e The total final energy in each region is then calculated by multiplying the extrapolated final energy intensity by the projected GDP (PPP) in each time period. Next, the extrapolated shares are multiplied by the total final energy to identify final energy demand for each of the seven energy service demands used in MESSAGE. Finally, final energy is converted to useful energy in each region by using the average final-to-useful energy efficiencies used in the MESSAGE model for each model region (:ref:`spatial`). .. _tab-quantssp1: -.. table:: Convergence quantile and income for each variable and region for SSP1 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each quantity and region for SSP1 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | @@ -123,7 +123,7 @@ The total final energy in each region is then calculated by multiplying the extr .. _tab-quantssp2: -.. table:: Convergence quantile and income for each variable and region for SSP2 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each quantity and region for SSP2 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | @@ -206,7 +206,7 @@ The total final energy in each region is then calculated by multiplying the extr .. _tab-quantssp3: -.. table:: Convergence quantile and income for each variable and region for SSP3 (for region descriptions, see: :ref:`spatial`) +.. table:: Convergence quantile and income for each quantity and region for SSP3 (for region descriptions, see: :ref:`spatial`) +--------------------------------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+----------+ | | | | | | | | | | | | | diff --git a/source/energy/tech.rst b/source/energy/tech.rst index 4f4fd1e..cda7184 100755 --- a/source/energy/tech.rst +++ b/source/energy/tech.rst @@ -27,7 +27,7 @@ no acceleration of technology diffusion is possible. In response to this limitat (Keppo and Strubegger, 2010 :cite:`keppo_short_2010`). These allow faster technology diffusion at additional costs and therefore generate additional model flexibility while still reducing the flip-flop behavior and sudden penetration of technologies. -:numref:`fig-difconstraint` below illustrates the maximum technology growth starting at a level of 1 in year *t*=0 for a set of five diffusion constraints which jointly lead to a soft constraint. +:numref:`fig-difconstraint` below illustrates the maximum technology growth starting at a level of 1 in year *t* =0 for a set of five diffusion constraints which jointly lead to a soft constraint. .. _fig-difconstraint: .. figure:: /_static/diffusion_constraint_example.png