This week’s blog marks the start of our latest seminar series. Dr Jacek Pawlak opened proceedings this term with a great talk on understanding energy use as society changes when, where and how we go about our daily routine.
We are currently seeing many changes in the energy sector. One key area is the slow move away from the old ‘predict & provide’ model, meeting ongoing, unconstrained, demand by building new generation and distribution capacity. With a growing population and electrification of areas such as transport, business as usual could turn out to be very expensive.
Instead, the sector is exploring ‘demand management’ strategies, which seek to explore combinations of pricing signals and incentives to change how and when people use energy. Demand-responsive pricing, deployment of smart meters, localised generation and storage or energy micro-trading, e.g. via vehicle-to-grid (V2G) technologies and services are all reflection of that underlying shift. The end outcome, hopefully, is a more optimal use the existing infrastructure on the one hand, but also ability to match spells of higher demand to times of higher generation capacity from renewable sources, e.g. windy or sunny days.
Thankfully it’s not wholly new ground, we have seen some something similar happen in an adjacent (yet increasingly interlinked) sector: transport. Back in 1970s and 80s, the ‘predict & provide’ paradigm was the prevalent policy approach, of which a classical example was construction of M25, London’s orbital motorway. Officially opened in its entirety in 1986, M25 very quickly became saturated (read: jammed), despite what had been initially deemed a future-proof capacity.
By 1994, it was realised that the proposed additions of up to further six lanes along some sections would only generate more traffic and not solve the original problem (Wikipedia actually offers a concise summary). Thus the demand management paradigm was burnt into transport policy-making, nowadays reflected in measures such as congestion and road pricing, peak and off-peak tickets, and various other (dis)incentives.
Demand management approaches, at their core seek to conceptualise and understand the very nature of demand: why do people desire a particular resource and what are they willing to give up to have it. The resource can be virtually anything: bread, pizza, cinema tickets, cars, but also energy or road space. The demand management perspective asserts that by gaining such core understanding, one masters the ability to influence it. Conversely, doomed is the one who remains ignorant to the demand response. The M25 case, with all its good intentions, did not account for the fact that more convenient driving is effectively a lesser cost of travel, driving its demand up.
A particularly interesting adaptation of the consumer theory, a branch of microeconomics looking at, among other things, the origins of demand, came from Gary Becker in 1960s Becker, a Nobel Prize laureate in economics, went on to interpret an individual’s choice over how to allocate their time between income-yielding work and non-work time as a utility-maximisation problem.
In his framework, everyone makes a choice between working and being able to consume and not working, but having time for leisure. This is then subject to budget constraints (one can only consume as much as they earn) and time constraints (the total time spent on work and non-work can only be equal to 24 hours a day). The consequent model provides a description of what motivates people to work more or less.
Kenneth Train and Daniel McFadden, himself a Nobel prize in economics laureate extended Becker’s framework to incorporate transport decisions, and thus provided what became a key theoretical underpinning for understanding of the origins of travel demand.
Fast-forward, the last decade has been characterised by a number of profound and simultaneous disruptions to ways in which people travel and conduct activities. These changes include increasing private vehicle electrification, emergence and large-scale proliferation of ride- and vehicle-sharing services, advanced vehicle automation, or resurgence in popularity of active modes or micro-mobility services. In addition, remote (online) activity participation, including in mobile contexts (travel, public spaces) has spread beyond narrow, highly-skilled groups of society.
The energy sector is also affected by those disruptions. Transport has remained one of the largest drivers of energy consumption, especially from fossil fuels. Despite efforts to curb energy consumption and the consequent pollution from transport sector, even in places with stringent regulations and substantial policy push to reduce emissions, energy consumption from transport continues to be major and growing. Furthermore, the aforementioned emerging activity-travel behaviour patterns are associated with an unprecedented spatial and temporal flexibility in conducting their activities, implying for the energy demand profiles not only change, but also more volatility and possibly unpredictability.
Here is where my current research comes in; we argue that part of the answer may be in more sophisticated demand management approaches driven by agent-based modelling (ABM). The idea of ABM is simple: rather than perceive the demand, whether for energy or traffic, as a de-humanised traffic flow diagrams or time series of kilowatts observed at a sub-station, the analysis looks at the individual person choices of activities, and the consequent, or derived demand for travel or energy. The aforementioned contributions by Becker, Train and McFadden remain natural theoretical underpinnings for such approaches. Nevertheless, so far those theories and their subsequent adaptations and extensions, have not jointly captured the original transport considerations and the need for the energy sector to manage demand mentioned at the beginning of this post.
In response, we propose a microeconomic framework, termed the HOT model (Home, Out-of-home,Travel) grounded in the Train and McFadden’s Goods-leisure framework, but suitably extended to incorporate the aforementioned emerging activity-travel behaviour patterns and their energy consumption implications.
In particular, the HOT model:
- Allows for existence of modes of transport in which the user can work productively or spend time enjoyably, e.g. train, bus but also connected and autonomous vehicles.
- Recognises the ability to work from home as well as undertake secondary work, both of which reflect remote (online) participation but also the gig economy.
- Understands the differential productivity associated with working at home and while travelling, as compared to typical office conditions.
- Includes an energy expenditure function that links allocation of time to various activities to energy consumption, and subsequently to cost to the budget and reduced consumption.
We have been able to show how the model can be operationalised, using semi-synthetic time use and consumption data from the UK, and allow predictions of demand side responses to transport and energy policies. Our proposed model can then aid in addressing incoherence within agent-based microsimulation models that typically lack the underlying and unifying behavioural framework to ensure a consistent (across the models) representations of the user behaviour.
The work remains an ongoing effort as part of the IDLES (Integrated Development of Low-Carbon Energy Systems) programme and we are currently testing the model in an empirical context of the data from the University of Oxford’s METER project. Through the collaboration, we hope to obtain a more realistic tool to accurately understand and forecast how demand for energy responds to pricing signals and incentives in an era of changing lifestyles and emerging activity-travel patterns.
Dr Jacek Pawlak
His principal interests are in modelling the interactions between Information and Communication Technologies (ICT) and activity-travel behaviour.
Jacek is currently involved in the EPSRC-funded Integrated Development of Low-carbon Energy Systems (IDLES) Programme at Imperial, where he is focused on Project 3: Data-driven models and decentralised control.