I’ve been asked a few times now to provide an account of the energy transition modelling featured on Beyond this Brief Anomaly over the past year or so, that goes beyond the very brief article for The Conversation in May, but that is more accessible than the detailed documentation provided in earlier posts here, here and here. The article presented here is intended to fill that gap. It’s based on the presentation I gave in July at a University of Melbourne Carlton Connect Initiative event on energy transitions, discussed in the introduction to this earlier post. The presentation abstract will serve for orientation:
Energy transition discourse in both the public and academic spheres can be characterised by strong and often fixed views about the prospects for particular pathways. Given the unprecedented scale and complexity of the transition task facing humanity, greater circumspection may help ensure collective efforts are effective. While significant attention has been given to the question of how to satisfy future energy demand with renewable sources, dynamic effects during the transition period have received far less attention. Net energy considerations have particular relevance here. Exploratory modelling indicates that such considerations are relevant for more comprehensive feasibility assessment of renewable energy transition pathways. Moreover, this suggests there may be value in asking broader questions about how to ensure energy transition learning and praxis is sufficiently ‘fit for purpose’.
Introduction and study context
In recent years I’ve noted an increasing tendency for energy transition discourse to be characterised by strong and often fixed views about the feasibility of various pathways. It’s this situation that forms the background context for the transition modelling that I’ll introduce shortly. In setting out the rationale for his new book Our Renewable Future , Richard Heinberg of the Post Carbon Institute recently summed up the situation as follows:
As just about everyone knows, there are gaping chasms separating the worldviews of fossil fuel promoters, nuclear power advocates, and renewable energy supporters. But crucially, even among those who disdain fossils and nukes, there is a seemingly unbridgeable gulf between those who say that solar and wind power have unstoppable momentum and will eventually bring with them lower energy prices and millions of jobs, and those who say these intermittent energy sources are inherently incapable of sustaining modern industrial societies and can make headway only with massive government subsidies.[source]
The unbridgeable gap Heinberg refers to is between people who see renewable energy, one way or another, as the best bet for our energy futures. Open the field out to include rusted-on fossil fuel adherents, or nuclear enthusiasts, and the polarisation is more extreme again.
At a time when we humans need to be moving towards a greater willingness to work together, this for me is a source of significant concern. Rigid positions can carry high stakes, especially if their foundations turn out to be weaker than anticipated. Humanity is now in a critical zone for the global carbon budget. Kevin Anderson from the Tyndall Centre for Climate Research in the UK describes the situation in the clearest possible terms: ‘low carbon energy supply can’t be built in time for 2°C.’ We are now at a point where supply-side energy transition responses, of any shape, can’t alone be relied upon to deliver the required rate of emission reduction.
Given the unprecedented nature and scale of the task before us, it seems fair to ask whether established methods for assessing transition feasibility tell us everything we need to know. Graham Palmer’s hierarchy of renewable energy plans (Figure 1) provides a very useful reference point for thinking about this. He points out that most feasibility assessments consist of energy balances for the post-transition state. These focus on what he calls the ‘simulation layer’, and involve the highest possible level of abstraction.The hierarchy identifies two additional layers of issues that need to be considered. Ronan Bolton’s work (presented immediately prior to the talk on which this article is based) could perhaps be seen as relating to the very important ‘first order layer’. This layer includes a wide range of practical issues that must be grappled with in actually implementing the simulation layer’s more abstract visions.
The ‘second order layer’ adds a further level of detail relating to important social, environmental and economic questions. This includes issues related to embodied energy and energy return on investment. Graham Palmer’s own research at the Australian-German Climate and Energy College is addressing this area, by developing ways to make energy return on investment indices more effective decision making tools.
The energy transition modelling exercise on which this article is based takes the second order layer as its general starting point, but extends this to include two further ideas:
- From the point of view of maintaining functioning societies during a major energy source transition, power return on investment may be a higher priority issue than life-cycle energy return. (This distinction between power return and energy return may benefit from further explanation. I’ve previously written about power return on investment here and here. I address this further towards the end of the article.)
- We need to look closely at what happens in getting from the present situation to the future situation, rather than focusing just on what the future might look like.
With these departure points in mind, the work discussed here explores the question ‘What happens to global energy supply during the transition phase, when we treat this as a dynamic process of change from the current state to a plausible future state, and when we take into account the significant additional energy demands associated with a major energy transition?’
Model background and design assumptions
Three broad principles shape the approach taken to exploring this question. These are worth keeping in mind when considering the model’s conceptual design and detailed implementation.
- Tracking the full implications of net energy requires a sufficiently comprehensive global view, where there’s no possibility of leaving out hidden energy subsidies through off-shoring effects. And related to this, it’s necessary to take a sufficiently integrated view that can deal with the effective cross-subsidy that a given energy source receives from other sources, and that reflects the quality differences between fuels and electricity.
- The approach I’ve taken is exploratory and isn’t intended to be the last word. The focus is on assessing the sensitivity of overall behaviour to various input parameters, in order to learn about the significance of dynamic effects for assessing the feasibility of transition pathways. This work isn’t about predicting a concrete future, and precision is therefore less important than making sure that the relevant structural relationships are accounted for.
- In the interest of supporting learning-oriented public conversation about energy transition, it’s important to explore the issue in a way that makes all structural assumptions and input values entirely transparent. To satisfy this principle, the model itself should be open access, and should allow anyone with sufficient interest to test how their own assumptions and preferences affect its behaviour.
The model developed to address these broad principles is represented schematically in Figure 2.
The model provides a physical characterisation of the global economy in terms of energy services in the form of work and heat. The global economy is represented as two sub-systems. First, an energy supply sub-system comprising each major energy source, with each separate source modelled as an independent stock of supply capacity.
The second sub-system is all other economic activity, where the output from the energy sub-system is used to produce goods and services for all purposes. The output from each energy source is converted to a mix of work and heat, via source-specific conversion efficiencies. The flow of work and heat from each source is then aggregated to give the total gross energy service supply.
For each source, a portion of this aggregated flow is then diverted back to cover requirements for new capacity emplacement, and for operation and maintenance. This leaves a net flow of energy services to cover all other economic activity. The central model logic involves attempting to maintain the net flow of energy services at a constant level as the contributions from each source change. The change process is driven by the gradual retirement of fossil fuel supply capacity, and this is replaced primarily by renewable sources.
The model is implemented using the Insight Maker system dynamics platform, which, while fairly basic in functionality, is entirely web-based. It is publicly available here, requiring nothing more than an internet connection and a web browser for access. The model actually runs in the browser itself. A screenshot is shown in Figure 3.
Approximately 30 of the input parameters are user-adjustable, directly from the browser interface (see sliders and text boxes at right of Figure 3). Anyone can clone the model and make their own modifications to the structure, simply by creating an Insight Maker account.
There are four basic structural assumptions built into this version of the model that need to be kept in mind in looking at the simulation results shortly:
- The transition involves a phase out of all fossil fuel energy supply over roughly 50 years. So we’re looking here at a much more modestly paced transition than what would be needed in order to move to zero carbon energy supply by 2035, for a chance of staying below a 2oC increase in global temperature relative to the pre-industrial baseline;
- During this period biomass, hydro and nuclear each roughly double in size;
- The model uses onshore wind and utility-scale PV to fill the gap in services as fossil fuel contributions decline; and
- Lithium ion battery buffering is used as a proxy for intermittency compensation, without implying that this is necessarily practical or preferable at this scale. It is, however, an increasingly popular view in the public imagination, and its implications are worth exploring in that context.
There are also several significant energy cost exclusions. Including these on the energy sector side of the ledger would further reduce the services available for other economic activity. These include:
- Adjustments to transmission and distribution infrastructure;
- Prime movers such as electric vehicles and other final energy use converters that are required as the proportion of electricity in the final energy mix increases; and
- Redesign and replacement of industrial processes, such as metals production, that are currently dependent on direct use of fossil fuels, both as energy sources and reductants.
Model simulation results and findings
With the foregoing introduction to establish context, we’re now set to look at the actual simulation results, with the model’s reference parameter settings. The chart for total final energy services shown in Figure 4 tells the overall transition story.
An explanatory note on the time scale (horizontal axis) may be helpful here. This is shown in units of ‘model simulation time’, with 0 years corresponding with 1900 in calendar time, 100 years with 2000 and 115 years with 2015. The period from 1900 until the start of the global energy transition in 2015 is simply the establishment phase for setting the global energy supply mix. The contributions from each source are very close to the actual values by the time the transition commences, but the trajectories from 1900 to 2015 are intended only to roughly approximate historical behaviour, with the exception of wind and PV. As will become apparent when we look at Figure 5, actual wind and PV emplacement rates (for all sources, including onshore and offshore wind, and rooftop and utility scale PV) are shown from the commencement of significant roll-out in the mid-1990s, until 2015.
Figure 4 shows total final energy services building to a peak between 2000 and 2015. At this point, the effect of net fossil fuel retirement kicks in, and the transition period commences. The net peak is around 6200 GW, and the gross peak around 6600 GW, with the difference between the two being the work and heat used by the energy sector. The region of primary interest here is the 50-year transition period between 2015 and 2065, indicated by the red bar under the time scale.
Of particular note is the decline in net energy services available to the economy between the start of the transition period in 2015, and 2035. For the reference parameter set, the minimum is just under 5500 GW, or around 12 percent below the pre-transition maximum. One of the most notable findings from the modelling exercise is that while the size of this trough is reduced by assuming more favourable values for relevant input parameters, it is never completely eliminated. This is the case even for significantly higher wind and PV emplacement rates. So a decline in net energy services available to the economy, in the early stage of the transition period, is a basic structural feature.
This decline results from the combined effect of two important issues. Firstly, we can expect emplacement accelerations for wind and PV to encounter practical limits. Figure 5 shows the changes in emplacement rates with time. As can be seen here, for the default model run emplacement accelerations are of a similar order to actual historical performance.
The model uses feed back control to increase the emplacement rates. In principal, feed forward control could be used to increase the rates earlier, in anticipation of the fossil fuel decline. But we’d still face the question of how much faster these could be expected to increase in practice. Here, the rate doubles in around 6 years for PV and a little longer for wind. But each of these then needs to go through two more doublings in less than 15 years. For reference, Figure 6 shows a stacked plot of the emplacement rates for all sources. As this indicates, the combined emplacement rates for wind and PV fairly quickly outstrip the historical rates for all other sources.
Figure 7 shows how the installed capacity builds over time. The emplacement rates that we’ve just seen, in addition to more modest rate increases for biomass, hydro and nuclear, are sufficient to maintain overall supply capacity at a constant level for around 10 years. After this, the growth in the contributions from wind and PV results in overall installed capacity building very rapidly.
But now, when we compare the installed capacity with the stacked plot for net energy services (see Figure 8), the second important issue that contributes to the decline shown in Figure 4 becomes apparent. We can see here how the net services from wind and PV grow much more slowly than the installed capacity.
This implies that, especially for sources requiring relatively high up-front energy investment, high acceleration in the emplacement rate acts as a brake on the rate of energy return. During the early stage of the transition, pushing the emplacement rate harder results in diminishing marginal returns on effort. This is particularly significant for PV, but it’s an essential feature of any rapid turnover in energy supply.
Figure 9 provides an indication of the energy sector’s size relative to the rest of the economy. It’s notable that while the situation depicted in the previous charts involves a doubling in energy sector size (the energy services ratio depicted in Figure 9 declines from a maximum of roughly 18 in 2000, to a minimum a little below 9 in 2047), in absolute terms this is not extreme. In fact, when the trough shown in Figure 4 bottoms out around 2035, growth in the energy sector’s service demands accounts directly for only a small fraction of the total shortfall.
It’s worth noting also that this takes into account a significant increase in overall conversion efficiency from final energy to work and heat. This is mainly associated with the increased proportion of electricity in the aggregate energy supply, but it also takes into account general efficiency improvements. The stacked plot of gross power output for all sources shown in Figure 10 indicates the overall effect of the efficiency increase, with around 25 percent less gross output required to deliver the same quantity of services post-transition compared with pre-transition.
The change in mean conversion efficiency for energy to services is depicted in Figure 11. The green line shows the mean efficiency across all sources for final energy to work and heat increasing from roughly 55 percent to just under 75 percent. Over the same period, efficiency for primary energy to work and heat increases from 40 to 60 percent.
Finally, Figure 12 illustrates what is perhaps the most illuminating finding from the whole exercise. Here, what I’ve referred to as ‘dynamic energy return on investment’ is calculated using the annual output from the total stock of generators for wind or PV, and the total energy expenditure for each source. The time series for this index shows it starting at zero and then increasing to the nominal life-cycle energy return on investment for a single generator only at the end of the transition period, once the overall supply stock has reached steady state. We see the same effect regardless of the life-cycle EROI value. In effect, what we’re seeing here is that the short-term power return on investment may be a significant transition rate constraint, even where life-cycle energy return on investment is favourable.
This shift in focus to ‘power return’ recognises that societies are dependent for their continued functioning on sufficient instantaneous rates of energy supply (energy supplied per unit of time), as a separate consideration to the quantity of energy available from a given source over its operating life. While most transition discourse focuses on changes in the sources from which energy is made available for human use, less attention is given to how equivalent rates of energy use can be maintained across the course of the transition. Here it becomes important to think in terms of how much power supply capacity can be made available from a new energy source, for a given investment of power from currently available sources. If we attend only to life-cycle energy return for a source, then this more immediate issue will be obscured. Any energy source that has a higher upfront energy requirement than incumbent sources will by its nature have a lower power return on investment. But even energy sources with relatively favourable upfront energy requirements can give low power return on investment during an unusually rapid turnover of the supply stock.
This insight is a direct result of shifting from a static to a dynamic view of energy transition. This alone could be viewed as proving the case for the importance of the dynamic view. It suggests fairly strongly why net energy considerations should be taken into account in assessing the feasibility of transition pathways.
While there’s still much scope for taking this analysis further, the findings to date point towards a number of broad conclusions that are likely to be fairly robust. The first of these is that the transition task ahead of us is likely to be more challenging than is popularly imagined. Replacing plant and infrastructure will take longer and have wider economic ramifications than is typically suggested by feasibility assessments that focus only on the ‘simulation layer’ in Graham Palmer’s hierarchy of renewable energy plans. It is not simply a matter of swapping out some existing plant and equipment.
In addition to this, the findings from the study support Kevin Anderson’s contention that we’re faced with a mismatch between the decarbonisation rate necessary to reduce the odds of dangerous climate change, and the practically realisable rate of transition to zero-carbon energy supply.
Finally, and I think most significantly, these two findings suggest a clear pathway for reaching beyond the apparent dilemma. This involves a much more serious engagement with demand-side questions, on the basis that any reduction in energy demand reduces the weight that supply-side responses must carry, therefore making the transition task easier. This needs to go well beyond a default focus on technology-mediated efficiency gains. The nature and origin of rich-world societal expectations for energy services, and related questions of cultural priorities, will need to become a far more prominent focus in energy transition efforts. This is not just a task that can be left to political and technical elites. It is a world-shaping endeavour that will affect every global citizen, and for which every one of us should be involved in equal measure.
I close then with a proposition that, to my mind, follows closely from this last point. The proposition is that the most important transition challenge facing humanity may not be about energy supply at all, but about opening our imaginations to new ways of learning together and to different visions of human success. That a transition from fossil fuels to renewable energy sources presents humanity with unprecedented social, political and technical challenges can be approached as a great opportunity. In fact, what is perhaps most remarkable here is that it should be necessary to highlight this at all. As a species, we are perhaps at our best when called to confront major existential challenges. If history is a reliable guide here, what matters is that the challenges with which we’re confronted present opportunities to find meaning and to learn together in our common endeavour. If the situation we now find ourselves in is not typically viewed in such a light, then learning to see it with fresh eyes may open new and altogether more rewarding pathways through the energy transition landscape.
 Heinberg, Richard, and David Fridley. 2016. Our Renewable Future: Laying the Path for 100% Clean Energy. Washington: Island Press. Online version available at http://ourrenewablefuture.org/.
 Anderson, Kevin. ‘The emission case for a radical plan’, presentation for the Radical Emissions Reduction Conference, Royal Society, London, December 2013. Viewed 1 July 2016 at http://www.tyndall.ac.uk/sites/default/files/anderson_-_radical_plan_conf.pdf.
 Bolton, Ronan. 2016. ‘Electricity market reform – a shared complex challenge’, presentation for EU Perspectives on the Energy Transition, Carlton Connect Initiative, University of Melbourne, 13 July. Presentation slides available here.