Building climate resilience into power systems plans: Reflections on potential ways forward for Bangladesh

The Electricity Journal 29 (2016) 32–41

Contents lists available at ScienceDirect

The Electricity Journal journal homepage: www.elsevier.com/locate/electr

Building climate resilience into power systems plans: Reflections on potential ways forward for Bangladesh D. Chattopadhyaya,* , E. Spyroub , N. Mukhic , M. Baziliand , A. Vogt-Schilbc a

University of Queensland, Australia Johns Hopkins University, United States World Bank, United States d Royal Institute of Technology of Sweden, Cambridge University’s Energy Policy Research Group, Sweden b c

A R T I C L E I N F O

Article history: Available online xxx

Keywords: Power systems planning Climate change Climate resilience Optimization model Stochastic programming Robust decision making

A B S T R A C T

The consideration of climate resilience in power system planning and operations by utilities around the world is very limited to date. This article assimilates some of the initial thoughts developed as part of a World Bank project on climate resilience for Bangladesh. It briefly reflects on the current literature, and focuses on the specific flooding risks faced in Bangladesh to illuminate the way forward to enhance planning practices. ã 2016 Elsevier Inc. All rights reserved.

1. Introduction While there is a rich global conversation on climate change and the need for building resilience in infrastructure, there has been surprisingly little change in the way power systems are planned and operated in response to the changing climate. Rather, the typical impact on the power planning community tends to be ad hoc response to a major event like an extreme heat wave, a hurricane, or tsunami.1 A conspicuous lack of evolution in planning and operations is particularly apparent in developing countries that continue to rely on deterministic least-cost planning methodologies with little or no consideration of some of the climate-change-related risks even when these thwart multiple projects or have led to common mode failures. Power system master plans developed using such a methodology in many cases do not consider risks around current weather variability (e.g., rainfall seasonality, flooding, heat waves, etc.), not to mention the additional variability due to climate change, which includes increased intensity and frequency of weather extremes such as heavy precipitation that impact flooding risk and water availability for cooling, and extreme heat waves that impact peak demand. In addition to the extreme events that have a

* Corresponding author. E-mail address: [email protected] (D. Chattopadhyay). 1 Nye (2010) provides a fascinating account of the technical as well as social and cultural impacts of power system failures. http://dx.doi.org/10.1016/j.tej.2016.08.007 1040-6190/ã 2016 Elsevier Inc. All rights reserved.

sudden impact on the power system performance, it is also important to consider slow onset events that are induced by climate change, e.g., increasing temperatures, sea level rise, salinization, etc. Such slow onset events often result in persistent poor performance and inadequacy of the system. Although sitespecific information is taken into account for renewable resources such as hydro, wind, and solar, historic data is used rather than projections for these resources that incorporate future climatic conditions. Operational profiles are thus drafted based on historical information on river flows, wind velocities, and radiation. Historical data on storm records, for example, are typically used for design of transmission towers and wind turbines. In countries or systems where risks associated with climate change are already taking their toll, the significant financial or even human losses, and the magnitude of investment required to strengthen the system against future risks, make it clear that there has to be a more systematic way to screen projects at a system level. Additionally, the observed frequency of extreme events in recent years has raised concerns among planners about the validity of the probabilities of extreme events that they now rely upon. Anecdotal evidence suggest that even developed countries seem to learn the hard way, via crisis. As an example, heat waves in Australia have been on the rise, and eventually took 173 lives in a disastrous bush fire before regulatory changes were implemented

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

to force the use of underground cables in sensitive zones.2 Likewise, better design of wind turbines is being put in place to withstand storms after they have been damaged severely in different parts of the world. Enhancement of power system resilience can be achieved through multiple solutions (U.S. Department of Energy, n.d.). However, the list of options is richer and probably less expensive at the early planning stage, rather than after the infrastructure is put in place. For example, the location of a power plant is fixed after the planning stage, and any re-design to make it more resilient – say, by a change of cooling technology or elevation of critical assets – will probably be costlier or even infeasible after construction commences. As a result, the electricity industry is beginning to recognize the need for development of methodologies that can assist it in building resilience into the system (WBCSD electric utilities, 2014). Regulators have also started considering updates on related standards and laws. Using climate-aware and climateresilient power systems planning practices could likely help save money, improve service delivery, and even save lives. The opportunity to build a resilient system instead of retrofitting it is more pronounced in developing countries, where significant investments on new assets are planned for the near future. In this article we use the case of Bangladesh to illustrate these points. It is a country that urgently needs to increase its generation capacity from 11 GW today to 57 GW in the next 25 years to meet demand growing at an expected annual growth rate of 6%. With limited remaining reserves of domestic natural gas, and no other significant primary energy resource in the country (including hydro), the major option for generation expansion is likely to be in the form of imported coal or liquefied natural gas (LNG). The Bangladesh government aspires to build 24 GW of coalfired power plants by 2022 (Reuters, 2016) and multiple LNG terminals that will allow for the import of natural gas (Senior Correspondent, 2016). Currently, only one small coal-fired power plant (250 MW) operates in Bangladesh using domestic coal. The latest Power System Master Plan supported the government’s aspiration to build several mega coal-fired power stations including 12 GW capacity in coastal areas that would run on imported coal (Japan International Cooperation Agency Tokyo Electric Power Company, 2016). There is, however, very little consideration of flooding risks—an issue that should be given more prominence as we highlight in a later part of this article. Given that Bangladesh is one of the most vulnerable countries to climate change, discussion and illustration of its power system vulnerability to climate change can indeed be found in recent research articles too—for instance I Khan et al. (2012), and Shahid (2012). The major risks identified in Iftekhar Khan et al. (2013) relate to inundation due to sea level rise, fluvial flooding, salinity or/and unavailability of cooling water, as well as rising temperatures. Despite the identification of the risks in the literature, the risks briefly discussed in the power system planning documents and are not explicitly included in the planning analysis to date. However, the climate risks are affecting the individual project site design choices as can be seen in feasibility studies. It is, however, not difficult to see why climate resilience has not been considered as a top priority in developing countries. There are many existing challenges that utilities, policymakers, and investors, including development partners, face in these countries. There is a modest but growing recognition of these issues in the latter countries as we briefly discuss in Section 2. However, for certain countries such as Bangladesh the exposure to climate events is of a scale that makes updating the power system planning process imperative, as we illustrate in Section 3. In Section 4, we

2

https://en.wikipedia.org/wiki/Black_Saturday_bushfires.

33

discuss how the power system planning model could be updated to be climate aware, and in Section 5 we conclude with recommendations on measures that can be adopted by policymakers and regulatory bodies to formalize a recognition of climate resilience. 2. Impact of climate change on power systems Resilience of energy systems, and power systems more specifically, to climate change is a relatively new area for research. A recent review can be found in (Panteli and Mancarella, 2015). The resilience of energy systems has been addressed, however, over a much longer period (see, for instance, Arghandeh et al., 2016; Espinoza, et al., 2016; Hamilton et al., 2016; Maryono et al., 2016; Molyneaux et al., 2016; Mukhopadhyay and Hastak, 2016). The climate change resilience issues have also been considered in other sectors like water, agriculture, and forestry (see e.g., Ching, 2016; Ho et al., 2016). The urban and transport sectors also have a growing body of work (See e.g., Bahadur and Tanner, 2014; Friend et al., 2014; Kernaghan and da Silva, 2014; Kiel et al., 2016; Meerow et al., 2016). A wider discussion about the policy issues surrounding resilience can be found in Chmutina et al. (2016). Additionally, the InternationalEnergy Agency has a dedicated website and material on this issue.3 Power systems operation (as opposed to planning) has always been closely interlinked with weather conditions and susceptible to extreme weather events that may in some cases be a large, if not the largest, contingency event. It is useful to clarify that the “climate change impacts” we are discussing relate solely to how this interdependence and susceptibility are likely to change over the years. The critical issue arising from climate change is that these natural hazards are projected to intensify, become more frequent, and become more unpredictable, as noted by Ebinger and Vergara (2011a,b), Mideksa and Kallbekken (2010) and Schaeffer et al. (2012), among others. In particular, issues that hold significant implications for power systems planning and operation include:  Increasing air and water temperatures;  Changing (and uncertain) precipitation patterns at the seasonal, decadal, and multi-decadal levels;  Changes in river flows (glacial melting, precipitation, etc.)  Increasing intensity and frequency of heat waves, storm events, flooding;  Sea-level rise;  Changes in wind patterns and intensity; and  Changes in insolation (solar radiation levels and patterns). Each of these factors on its own, and in several cases in combination, may lead to systemwide disruptions, including common mode failure of a large part of generation. For instance, increased air temperature would tend to increase demand, reduce thermal conversion efficiency (and thus firm capacity), diminish transmission capacity, and increase the temperature of water for cooling—the collective impact of these events can, in extreme, cases may be a catastrophic event for the power system. Depending on the geography, this may well account for the largest power system contingency and hence an issue that planning should recognize. Mechanisms describing the relationship between climate change and power system performance are well known for

3 http://www.iea.org/topics/climatechange/subtopics/resilience/. The U.S government has a partnership related to climate resiliency here: http://energy.gov/ epsa/partnership-energy-sector-climate-resilience, and a useful toolkit here: https://toolkit.climate.gov/topics/energy-supply-and-use.

34

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

available technologies—at least qualitatively. In brief, the impacts on thermal power plants (coal, natural gas, nuclear) are expected to be characterized by reduced efficiency and net capacity. Two reasons explain this: (1) increase in ambient temperature; and (2) reduction of thermal discharge potential to water resources because of lower water volumes and higher water temperatures (Rubbelke and Vogele, 2011), Linnerud et al., 2011).4 Hydropower is expected to be significantly affected by climate change, mainly because of change in hydrological cycles caused by temperature and precipitation change and glacier melt (Mukheibir, 2013). Transmission and distribution lines are also sensitive to ambient temperature increase since thermal limits, loss factors, and sagging of lines, depend on temperature (Kezunovic et al., 2008). Renewable technologies such as wind and solar are also vulnerable to climate change because of their high dependency on nature (Sailor et al., 2008; Tobin et al., 2015; Cradden et al., 2012). For example, solar is another resource sensitive to climate change through insolation levels and patterns, cloud factor, and temperature (Jerez et al., 2015; Crook et al., 2011). Climate change is also expected to increase the frequency and severity of extreme weather events such as heat waves, flooding, and ice storms (Melillo et al., 2014). Extreme events stress the limits of the power system, and can lead to disruptions of the power supply as recent experience in the U.S. reveals (Edison Electric Institute, 2014). They might lead to failure of transmission system components or damage of towers, substations, and conductors e.g., under extreme wind gusts (Rezaei et al., 2016), during wildfires, caused by long droughts (Kezunovic et al., 2008) or during floods (Panteli and Mancarella, 2015). Extreme events also affect generation. For example, flooding threatens reservoir operations (Mukheibir, 2013) and power plant equipment (Espinoza et al., 2016). Extreme wind along with icing might lead to damage of wind turbines (Chou and Tu, 2011). Extreme hail could also damage photovoltaic panels (Patt et al., 2013). While extreme temperatures observed during heat waves might affect the cooling of the power plants (The Guardian, 2003). In the past, most studies focused on illustration of the impact of climate change on the operation of a pre-defined power system. For instance, van Vliet et al. (2012) explore the potential decrease of available thermal capacity due to water levels and temperature for an assumed thermal fleet in the U.S. and Europe under two climate change scenarios. However, more recently, studies have been using capacity expansion planning models to illustrate how the optimal mix might change in the presence of climate change. Jaglom et al. (2014) explore the impact of temperature under different climate scenarios for the American power system using ICF’s planning tool (IPM). Parkinson and Djilali, (2015) model the impact of temperature and streamflow changes for the electricity system in British Columbia. Cohen et al., (2014) enhances NREL’s planning model (ReEDS) by incorporating water availability constraints. Although there are a growing number of studies examining the impact of climate change on power systems looking at individual impacts, to the best of our knowledge there is no comprehensive framework for their consideration that takes into account their interdependencies.5 However, actual cases demonstrate that one extreme event, such as the prolonged drought in Australia, affected not only hydro production but also availability of cooling water that eventually led to shut down of several major coal-fired power

4 We briefly discuss the relationship between climate change parameters and power system. However, we refer the reader to the following resources that describe the relationships in depth (Chandramowli and Felder, 2014), (Mendiluce, 2014), (Ebinger and Vergara, 2011a,b), (Wilbanks et al., 2013). 5 The U.S. DOE Toolkit mentioned previously goes some way in this regard.

stations in three states in March 2007. Wholesale electricity prices shot up in extreme days to close to AUD 7000 per MWh (AER, 2008).6 As another example, reductions in electricity production in the Western Balkans due to reduced river flow would be concurrent with an increase in cooling demand, which is projected to increase about 50% in a 4  C world (World Bank, 2014). Moreover, most studies continue to focus their modeling on the behavior of a specific power system component: e.g., ambient temperature on thermal capacity and efficiency of a specific power plant (Arrieta and Lora, 2005). More recent work on power system resiliency also attempts to model weather events to take into account common mode failure cases (Espinoza et al., 2016). However, most of the solutions examined in the wake of disaster events focuses on improving preparedness, through forecasting tools that will help utilities pre-position crews and replacement parts (Singhee et al., 2016), and reducing post-disaster recovery time through employment of easy to replace or temporary equipment (“ABB and U.S. policymakers meet to address grid vulnerability issues,” n.d.), (Hong Kong Business Environment Council, 2015). A review of the literature and industry practice generally points to the lack of a standard planning tool that takes into account climate data as well as associated uncertainty. This gap might be explained by three reasons, namely: 1. First, the nature of uncertainties is in itself a fundamental challenge that sets these problems apart from those dealt in conventional power systems planning. The degree of uncertainty attached to climate projections is quite high (“deep uncertainty”) and it might be perceived as a barrier to adaptation investments when the decision-maker is new in adaptation planning (Hanger et al., 2013)7 ; 2. Second, climate data and weather projections readily available require significant processing in order to be in a spatial and temporal granularity useful for decision-makers; and 3. Third, characterization of the stochastic performance of equipment under different hydro-met parameters is in an early stage, with limited data available. For example, there is anecdotal information, not tested for its statistical significance, that there might be a tipping point for ambient temperature beyond which probability of failure of distribution lines increases dramatically (Hammer et al., 2014). There are multiple degrees of freedom in power plant design and the characterization of vulnerability and performance that might differ even for the same type of power plant. EPRI, recognizing this flexibility, has started developing “SOAP” that might suggest optimal power plant design under different conditions, attach specific costs to that design and operational characteristics that can be part of the planning tool. All of the studies identified in this review do address qualitatively or quantitatively some aspects of climate change. However, none systematically attempt to decipher the impacts and allow planners to explore the options in full. 3. Discussion on Bangladesh

6 Australian Energy Regulator, State of the Energy Market, 2008. P.146. https:// www.aer.gov.au/system/files/State%20of%20the%20energy%20market%202008% 20complete%20report.pdf. 7 Note however that even in case of “traditional” uncertainties such as the ones related to fuel prices or demand growth, planners have been reluctant to implement stochastic planning tools. So in general, there is limited working knowledge of planning models that explicitly take into account uncertainty and planners have mainly resorted to a deterministic incorporation of uncertainty e.g., through planning reserve margins.

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

Bangladesh is located at the delta of three major rivers: the Ganges, the Brahmaputra, and the Meghna. Most of its land is low-lying, prone to flooding. The floods in Bangladesh are divided into monsoon river flood, flash flood, local rainfall flood, and storm surge flood (Bangladesh Water Development Board (BWDB)). Projections on changes in flooding patterns due to climate change have attracted interest from research community given the devastating impact of past floods and the low lying profile of the country ((Carrol et al., 2011), Ali, n.d.). While climate change will have more effects than change in flooding patterns in Bangladesh, we will focus on this aspect for time being given its importance for site selection and consequently power system planning. Protection against flooding has been an important consideration for power plant construction in Bangladesh, as cases reported in Table 1 indicate. Developers have designed plants so that they are protected against the 200-year highest flood level or the highest historical record. In cases such as Matarbari significant construction work is needed to raise the elevation of the power plant by heights as much as 9 m. Climate change introduces a challenge for power system developer and power system owners since the 100-year highest flood level (and 200-year highest flood level, respectively) is projected to change under different climate change scenarios. Fig. 1 shows the uncertainty a planner would face if she was about to use the readily available riverine flooding data provided in WRI to decide on the flood protection measures required (for the power plant development of approximately 23 GW capacity).8 If climate projections from NonESM1-M were used, the developer would not have to build any flood protection for approximately 75% of the proposed capacity while if data from IPSL-CM5A-LR were about to be used, the required percent decreases from 75% to 58%. Similarly, we observe that only 5% of the proposed capacity requires flood protection higher than 6 m in the case of NorESM1-M but this percent increases to 32% when data from IPSL-CM5A-LR are used. Consideration of historical coastal flooding data due to storm (Japan International Cooperation Agency (JICA) Tokyo Electric Power Services Co, 2015) suggests that even more candidate coal capacity is exposed to 200-year maximum flood depth. Therefore, significant construction work will be required to protect the power plant. Only 8 GW of the 25 GW considered in Fig. 2 fall in area that is not vulnerable to flooding when return period of 100 years is considered.9 Data on the cost of construction of the embankment to raise the site are only provided for the power plant in Khulna. There, site leveling is estimated at USD 24 million, which is less than 2% of the total plant cost. However, even that cost estimate seems to be outdated and probably not accurate since the overall project cost has increased from USD 1.7 billion in the feasibility study (NTPC Limited, 2012) to at least USD 2 billion, according to recent news coverage (“Rampal power plant deal signed,” 2016). Assuming price for the filling material at USD 13/cubic meter (domestic price estimate from (Japan International Cooperation Agency (JICA) Tokyo Electric Power Services Co, 2015)) construction cost for Matarbari would be higher by USD 221 million compared to a similar power plant that does not require any flood protection. Those estimates indicate that the construction cost to harden the power plant infrastructure against flood risk might be significant and planners should try to consider them when prioritization of sites is considered among different candidates. Hence, as a first step, planners should try to incorporate data into

8 Using approximate location for this 23 GW capacity based on best available information in the current planning documents. 9 Maximum flood depth projected by any of the 4 riverine flooding model and the historical coastal flooding data.

35

the planning process related to the cost of flood protection and the damage that the system might suffer in case of inundation.10 Those data would be adequate to decide on an optimal protection level per plant and estimate the investment cost required. However, as a second step, planners face the uncertainty among the different climate projections. Incorporating such pervasive uncertainties is paramount and necessitate better planning methodologies—an issue that we discuss in the next section. 4. Methodology for climate-resilient power systems planning Power systems planning models have evolved over the years to consider a range of criteria, starting with reliability and followed by costs, energy security, and environmental constraints, among others. However, the methodology in practice has by and large been focused around a deterministic least-cost planning model. Uncertainties including those surround load growth and fuel prices have played a significant role, but have largely been addressed through looking at specific scenarios around these parameters. There are indeed stochastic power system planning models including the seminal works on stochastic programming developed in the 1980s (Birge and Lovueaux, 2011), among other techniques. Climate resilience is part of this evolutionary process in planning, but clearly one that strengthens the need to shift towards probabilistic techniques that explicitly capture uncertainties. In this section, we have outlined key attributes of the methodology that can be overlaid on existing planning models and also touch upon data issues relevant to such analyses. In particular, we discuss two methodologies that have been proposed as relevant for adaptation decision-making: stochastic programming and robust decision making (OECD, 2015): 1. A multi-stage stochastic linear programming (SLP) model that treats climate change related and traditional uncertainties (e.g. load growth) to decide on an least-cost plan considering location, technology, and timing of new power plants; and 2. A robust decision-making (RDM) model following Lempert et al. (2003) principles that evaluates alternative strategies and suggests an adaptive strategy which perform well across a wide range of possible future outcomes. These two methodologies approach the same problem using largely the same input and even the underlying physical model of the power system, but with a subtle philosophical difference. SLP falls in the realm of more conventional optimization that looks for the lowest-cost outcome, through establishment of analytical relationships between climate parameters and power system parameters. SLP explicitly characterizes the uncertainty through probability assignment to different realizations of the power system parameters. As long as the probabilities of different scenarios have been provided by the planner, the process is reasonably “automated,” usually requires no intervention from the planner, and in many ways conforms to the conventional least-cost planning models. Robust decision-making, on the other hand, does require intervention by the planner to interpret the results, since at minimum a threshold has to be decided by the planner that determines if a particular outcome is satisfactory or not. In many ways, robust decision-making is more focused in selecting “good” solutions rather than the optimal outcome. The major attribute of robust decision-making that makes decision-makers prefer it for the climate change uncertainty problem is that it does not require the planner to specify probabilities of different futures. Moreover, in most practical applications many future states are explored as

10 However, there are other impacts of flooding such as disruptions to coal transportation, coal storage and mining that are not covered by these measures.

36

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

Table 1 Flood protection level standards for planned power projects in Bangladesh. Technology

Power Plant

Capacity (MW)

Elevation (m)

Filling required (1000 m3)

Criterion to decide on elevation level

Coal Natural gas Natural gas Natural gas Coal

Matarbari Bhola Meghnaghat Haripur Khulna

2400 220 450 360 1320

9 3.7 5.33–6.83 4.9 3.75–4.25

17,000 160 1880 200 8000

Previous highest record, statistical analysis13 200 year highest flood level +1 m14 200 year highest flood level +1 m15 Higher recorded water level of Sitalakhya river +1 m16 Area’s highest flood level17

Fig. 1. Coal plant capacity by flood depth for 100-year return period.

the first step and provide to the planner a good overview of the vulnerabilities and the weaknesses of system or strategies that the planner should act upon. Both of those methods are using or enhancing a traditional least cost optimization model that we summarize first before discussing the attributes of SLP and RDM. 4.1. Least-cost optimization model In this section, we summarize the attributes of a typical leastcost optimization model applied in most systems in one form or another, including developing countries. Bangladesh, for instance, has been doing power systems master planning exercise under the aegis of JICA since 2005 using a type of least-cost planning analysis.11 The planning work in 2010 and 2015 has primarily been carried out by the consulting arm of Tokyo Electric Power Company (TEPCO), funded by JICA. TEPCO’s generation planning methodology differs from planning tools such as ICF’s IPM and NREL’s ReEEDS since the optimal mix in terms of fuel composition is first decided based on load duration curve for the final year of the horizon. Then planner adds the relevant units to achieve this longterm optimal mix and the performance of alternative generation expansion plans characterized by different mix of coal and gas is simulated through a dispatch optimization analysis.12

11 JICA has completed two of these plans in 2005 and 2010 with the third one for 2015 currently in the final stage of development. Bangladesh Power Development Board that owns majority of the generation (primarily gas-fired power stations) is the primary recipient/consumer of the generation plan. TEPCO works closely with BPDB and other stakeholders in Bangladesh including the Power Grid Corporation of Bangladesh to develop five alternative generation planning scenarios with largely the mix of gas and coal varied across these. 12 For instance, the JICA presentation of the draft PSMP 2015 results to the World Bank in May 2016 shows the five scenarios for 2041 include the share of coal varying from 15% to 55% (in 10% steps) with commensurate decrease in share of gas from 55% down to 15%. The total share of coal and gas is 70% in all cases with 25% of the mix from other resources (including import and nuclear) and oil retaining a 5% share in the long term. JICA High-level Discussion on PSMP 2015 Presentation, April 2016.

However, the least-cost optimization model central to both methods minimizes total system costs including capital cost of building new plants, operating costs of existing and new plants, and the cost of unserved energy, subject to a set of physical constraints. The constraints considered in this first capacity expansion planning model applied for Bangladesh are provided below. However, as the planning exercise in Bangladesh becomes more regular, physical constraints representing the transmission network and the laws that govern its operation should also be incorporated:  Load balance constraints for each time period modeled,  Generation (maximum) limits,  Fuel constraints that limit the consumption of different type of fuels,  Budget constraints that reflect the limited available capital, and  Land use constraints that constraint the land that can be used for power plant development. The key input data for the model includes:  Cost: investment costs for generation expansion and fuel prices.  Load: Load forecasts in the form of load duration curves (LDC) for the planning years;  Generation: Operational characteristics of generation plants such as thermal efficiency, variable and fixed operation and maintenance costs. 4.2. Stochastic programming approach The deterministic least-cost planning model can be converted to a stochastic programming model by incorporating different

13 (Japan International Cooperation Agency (JICA) Tokyo Electric Power Services Co, 2015). 14 (Environmental Resource Management Pte. Ltd Lanco Power International Pte Ltd, 2012). 15 (ESG International, 2000). 16 (ESG International, 1999). 17 (NTPC Limited, 2012).

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

37

Fig. 2. Candidate coal capacity (MW).

scenarios that represent the uncertainties faced by the planner. One of the major uncertainties in Bangladesh relates to the demand growth, which is primarily induced by uncertainty over long-term economic growth and secondary by climate driven temperature sensitivity of load. We assume that demand uncertainty can be fully characterized through discrete scenarios for which the associated probabilities can be obtained. The problem may be formulated as a two-stage stochastic program in case the planner believes that there is a point in the future when the uncertainty will be revealed. In case this point exists, this is the border between the first and second stage. For socio-economic growth, this might be a fair assumption for a country with a stable economic outlook for a 10-year horizon. First-stage decisions (“here and now” decisions) require an investment in generation capacity under uncertain future conditions. Second-stage decisions (“recourse decisions”) consist of operational generation decisions and future capacity additions. These decisions are made deterministically once uncertainty has been revealed (when the actual demand trajectory has been realised), and are conditional on both the realized demand and the first-stage decisions. The second-stage decisions are scenariodependent. Fig. 1 gives a description of this two-stage model. A more realistic multi-stage formulation (where uncertainty is revealed at more than one stage in the planning horizon) may also be considered exactly in the same fashion, albeit at the expense of a larger more complex model. These scenario-based stochastic linear programs optimise under uncertain future conditions by producing contingent decisions over a number of future scenarios. The objective is to find an expansion plan that will be implemented before uncertainty is revealed, such that the expected cost of this plan over all scenarios is minimized. The stochastic linear program differs from the deterministic linear program in that parameters and objective coefficients might have multiple realizations over time, corresponding to the multiple scenarios active at a given

time. First stage decisions have only one realization, independent of the scenario realized. They can be interpreted as ‘here-and-now” decisions that must be taken when only probabilities of the different scenarios are known. However, Stage-2 or resource decisions are made after uncertainties are resolved and it is possible to react to the events that follow, e.g., build low or high depending on the demand growth paths. This is shown diagrammatically in Fig. 3, where X is the here and now expansion variable, and Y are the recourse variables. The SLP setup is suited for modeling a range of uncertainties including those pertaining to climate change. For instance, the change in capital costs to incorporate flooding risks, restrictions on plant availability, increase in peak demand due to heat waves, and changes in solar/wind resource quality – are all factors that can be modeled as alternative climate scenario paths. There are however challenges in implementing such an approach – for instance, it is difficult to get good quality data or to make any assumption about the evolution of uncertainty around these parameters in the future. Nevertheless, it is worthwhile making an effort to use limited data than to ignore the issue altogether. 4.3. Robust decision-making model The RDM methodology identifies strategies that will perform satisfactory across a wide range of possible futures and according to a set of different metrics for success. It involves the generation of a large set of scenarios exploring uncertainties among several dimensions, including both socio-economic uncertainties and climate change uncertainties. The performance of different strategies is evaluated across these scenarios using different metrics. This allows the identification of strategies that perform well over a large set of possible futures. The first step is to set up the problem using the XLRM framework, which stands for uncertainties (X), policy options or levers (L), relationships and models (R), and metrics for success (M).

Demand growth scenario Stage 1

Stage 2 YHigh

X

Uncertainty revealed

YBase

High, Base,

YLow Low,

Fig. 3. Compact formulation: defining scenario structure and handling non-anticipativity implicitly.

38

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

4.3.1. Uncertainties Economic and climate include:  Demand growth, future fuel prices, prices for imported electricity, available capital for investments, feasibility of energy efficiency improvements  Future yields by type of power plants, future availability of sites (both resulting from climate change) 4.3.2. Levers: options available to decision-makers Levers refer to the list of different options available to decisionmakers. They include the list of different generation power plants, interconnection options, demand-side management, etc. Levers also include a quantification of these options in terms of capital and maintenance cost, yield, fuel used, maximum output, etc. 4.3.3. Relationships: models The underlying least-cost physical model that determines the optimal power plant portfolio given an exogenous demand (that varies year by year, and within a year for a set of seasons), fuel prices, and a list of power plant options characterized by their capital and maintenance costs, yields, and fuel requirement provides an optimal (least cost investment plan), as captured in the least cost optimization model. 4.3.4. Metrics for success To measure the performance of portfolios, decision-makers might decide on different metrics. Examples of metrics might include levelized cost of supplied electricity, total unserved demand and total unserved reserves, and CO2 emissions. For each of these metrics, an acceptable threshold needs to be determined. For instance, an acceptable threshold for the levelized cost of electricity might be USD 0.20 per kWh. If a strategy yields a higher cost, it is deemed unacceptably expensive i.e., strategies that yield a higher cost may not be considered. The same approach is applied to other metrics. The second step is to generate a large set of scenarios (say, 200). Once the uncertainties and their respective ranges have been decided, Latin Hypercube Sampling can be used to generate a number of scenarios that covers a large part of the uncertainty. Then, the optimal investment plan is identified for each scenario by solving the least cost optimization model. This yields 200 optimal investment plan, each contingent to one specific scenario. Next, these 200 optimal investment plans may be grouped together in a smaller number (say, 20) of representative plans, considered at this point candidate strategies. The performance of each of these 20 strategies is then assessed in each of the 200 scenarios (for a total of 4000 model runs), according to the metrics for success and their threshold set up above. Finally, statistical analysis of different portfolios performance is used next to identify the more robust strategies, i.e., those which perform well over a large set of scenarios. Moreover, conditions under which the robust strategies fail can also be investigated, and alternative strategies can be identified that perform well under those conditions. This way, adaptive strategies can be developed, where the decision-makers first pursue a robust strategy, and adapt it over time if uncertainty over specific conditions evolves in the direction identified before. 4.4. Discussion on climate model data 4.4.1. Climate models and projections A range of climate variables can impact the power sector infrastructure. Projections of key climate variables (temperature and precipitation) are obtained from global climate models (GCMs) that are mathematical formulations of various processes in the

climate system (e.g. radiation, energy transfer by winds, cloud formation, evaporation and precipitation of water, and transport of heat by ocean currents). Typically, GCMs simulate the climate system’s response to CO2 emissions and other anthropogenic activities that impact the system in order to generate projections of climate variables such as temperature and precipitation. These output projections are based on certain assumptions of economic activity and emissions across sectors that contribute to a certain level of atmospheric CO2 concentrations over a time period, referred to as concentration pathways. Model calculations are made for a grid size ranging from 100 to 500 km (horizontal and vertical) covering the entire earth. Give the coarse resolution, these projections are downscaled for local analysis of climate impacts on infrastructure using either regional downscaling (using regional GCM) or statistical downscaling using historical observations that are corrected for biases. Downscaling further increases the uncertainty associated with the climate projections. 4.4.2. Climate variables and climate variable derivatives Climate variables to be included in a specific power system analysis should be chosen depending on the geographical location and the country’s current and future power infrastructure and demand. The variables will depend on the scope of the analysis as well, i.e. generation (power supply technologies and demand), transmission, and/or distribution. Climate variable projections (temperature and precipitation) are not always used directly in the power system planning model. For example, cooling water availability for the fossil-based units is more relevant than just precipitation projections, and cooling degree days (change in expected in power demand for cooling due to weather and climate) is a more relevant index than just extreme temperature projections. In summary, projections of climate variables usually require further processing to develop climate derivatives/indices to estimate their impact on variables used to characterize power supply and demand in the planning models. The complexity of climate derivatives will vary depending on the power system stage (generation, transmission, distribution), supply technology (fossil, solar, wind, hydro), and the types of climate impacts to be assessed (extreme heat events, flooding/drought risk, wind variability, change in irradiation, coastal storms, change in water availability for hydro, etc.). The next step is to incorporate a mathematical functional relationship between these derivatives and variables in the system planning model. These mathematical functions may be available in published literature for different countries or can be estimated for the specific country using empirical data. Some of the key climate variables relevant to the analysis include:  Temperature: Extreme temperatures can impact the generation efficiency, the overall demand, and peak electricity demand. Temperature projections for the country/region can be obtained from several GCM, typically at a resolution of 100  100 km. Several online portals and climate research groups provide projections from individual models and also ensemble of models (for specific climate scenarios, e.g. RCP 8.5, RCP 4.5). For example, the Climate Analysis Tool provides data from 17 GCMs developed for the Intergovernmental Panel on Climate Change Fifth Assessment Report (CMIP5 Archive) run across two representative concentration pathways (RCPs 4.5 and 8.5).  Riverine and coastal floods: Riverine and coastal floods can impact the site selection and viability of existing and proposed generation sites. To assess the potential flood risk of existing and proposed power plants in Bangladesh, we combined spatial data on flood hazard with the approximate location of power plants. The flood hazard data comes from separate global models

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

representing river and coastal floods. River floods are represented using the global model GLOFRIS, which produces estimates of inundation depths in decimeters for each grid cell at the 30 arc-second resolution (approximately 1 km x 1 km at the equator). Additional data sources are being explored for coastal flood analysis.  Wind and solar variability: Local resource assessments are done before selecting sites for wind and solar power generation. However, wind and solar variability in the long term due to climate change can impact the overall long-term viability of these investments. Data on wind speed and insolation can also be sourced from the GCMs that are being deployed to obtain the temperature data for the same climate scenarios (RSP 8.5 and RCP 4.5). However, the uncertainty associated with wind and insolation projections is usually much higher (than temperature and precipitation projections) given the assumptions built into the factors used for post-processing, e.g. cloud cover, seasonal and inter-annual variability etc.  Extreme wind and coastal storms: These present a significant risk for wind power generation sites, distribution lines and in many cases for the transmission systems as well. A spatial analysis of potential wind power locations and distribution lines can be performed assessing for risks of storms, cyclones, etc. The historical dataset for coastal wind (and coastal inundation) can be used for this along with projections (to the extent these are available). The probability of extreme wind events can be used to identify locations that are at a high/moderate risk of potential damage.  Precipitation and river runoff: Precipitation projections are also available through GCM models. These projections are further used as an input for river basin analysis using hydrological models to estimate river runoff for the hydropower generation site being assessed. Change in precipitation due to climate change is incorporated in the hydrological model to estimate change in expected runoff at the plant site that can then impact the design of the plant. 5. Concluding remarks As the state of the literature and discussion on methodology suggests, building in climate resilience in a power system plan can be a substantial task both conceptually and analytically. It is important that some of the implementation challenges and ground realities for developing countries are kept in mind to ensure a practical work program to incorporate climate resilience in the plan. In this concluding section, we have made a few comments on the way forward to implement for a climate-resilient power system plan keeping in view some of these issues for Bangladesh. It is envisaged that planning methodology will be enhanced by associating climate parameters with traditional power system planning model counterparts. To the best of our understanding, considerations of climate resilience are largely absent from the current planning method. Examining a wide array of power plant site choices can be a significant option to build resilience (against climate change) that exists only in the planning stage. However, this is rarely done to the best of our knowledge and certainly has not been considered in Bangladesh. Although generation projects in the financing stage go through a rigorous check on flooding standards, flooding risks are not included in the planning stage. While it may not be a material issue in most countries, it is a significant one in Bangladesh and getting worse in some parts of the country. A restricted number of sites and potentially significant dip in availability of plants due to flooding may limit supply and call for a wider range of resources, including power imports, that have yet to gain as much attention as domestic capacity expansion. On the other hand, peak demand and

39

especially extreme peak demand is also likely to escalate due to climate change. Bangladesh is a summer peaking country with peak load reasonably sensitive to temperature. If the temperaturesensitive air conditioning and cooling loads continue to rise rapidly over the next 25 years, a heat wave similar to the one experienced in April/May 2016 may cause the peak load to vary by 1500– 2000 MW (i.e., may make a difference of up to $2 billion in new peaking generation and/or demand response related investments—this is equivalent to 1.33% of the country’s GDP in 2013). There are other significant issues around cyclone/storm risks that can impact on a range of infrastructure including wind turbines. Again, the climate models by and large predict that extreme weather events such as heat waves and cyclones/storms are likely to be on the rise and stronger in tropical countries. There are at least three reasons why a fresh look at capacity planning in Bangladesh is important: 1. An acute need for new capacity in Bangladesh to increase it nearly five-fold over the next 25 years; 2. The potential for significant climate-change-related impacts, especially increased frequency and severity of floods as predicted by the climate models; and 3. All major scenarios considered to date focusing heavily on baseload domestic coal/gas projects leaving out cross-border power imports on an equal footing that have the potential to render the Bangladesh power system more resilient. For instance, it is not entirely outside the realms of possibility that flooding risks alone may increase significantly the construction cost for a good part of the 24 GW candidate coal capacity given the paucity of land sites and the known flood risks. On the other hand, options like power imports from India/Nepal/Bhutan and demand response that can potentially play a much bigger role than has been considered to date, could render the supply mix to be more diversified, less risky and climate-resilient. Bangladesh’s power sector strategy has in the past been somewhat unilaterally focused on one fuel (namely, gas), before a recognition that the resource (at least for domestic onshore gas) is limited started the dash for coal post-2010. Although coal development projects have been identified since 2011 when the PSMP2010 was published and despite the interest expressed by investors, no new coal capacity has been added to the system and none is projected to be added before 2019. Some of the obstacles identified relates to the timing and the time to approve the environmental impact assessment (EIA) along with additional costs identified through additional requirements after EIA’s approval. Moreover, public opposition hinders the development of coal given the pollution-emitting profile of coal and its major land requirements. A fresh look building on the TEPCO/JICA plan and incorporating some of the key climate change related risk may be worthwhile—if for no other reason than to ensure that the proposed coal/gas scenarios do not expose the country to an undue risk that may surface at the project analysis level, or worse, lead to stranding of assets worth several billion dollars that the country simply cannot afford. It is useful to build planning models that explicitly consider these risks to essentially isolate the impact flooding risk and other extreme weather events on the plan. If these impacts turn out to be material, it would be well worth exploring how the resilience issues can be addressed by re-prioritizing the projects and diversifying the generation/import mix.

References ABB and U.S. policymakers meet to address grid vulnerability issues. (n.d.). Retrieved from http://www.abb.com/cawp/seitp202/ c91d04b865ad629885257e4c004faaba.aspx.

40

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41

Ali, S. (n.d.). Effects of Climate Change on Floods of Bangladesh: Learning from the past. Arghandeh, R., von Meier, A., et al., 2016. On the definition of cyber-physical resilience in power systems. Renewable Sustainable Energy Rev. 58, 1060–1069. Arrieta, F.R.P., Lora, E.E.S., 2005. Influence of ambient temperature on combinedcycle power-plant performance. Appl. Energy 80 (3), 261–272. http://doi.org/ 10.1016/j.apenergy.2004.04.007. Bahadur, A.V., Tanner, T., 2014. Policy climates and climate policies: analysing the politics of building urban climate change resilience. Urban Clim. 7, 20–32. Bangladesh Water Development Board (BWDB). (n.d.). Flood Forecasting & Warning Centre. Retrieved March 8, 2016, from http://www.ffwc.gov.bd/index.php/ definitions. Carrol F., Christidis N., Fullwood J., de Gusmao D., Golding N., Good L., . . . Warren R. (2011). Climate: Observations, projections and impacts: Bangladesh. Retrieved from http://eprints.nottingham.ac.uk/2040/16/Kenya.pdf. Chandramowli, S., Felder, F.F., 2014. Climate change and power systems planning— opportunities and challenges. Electr. J. 27 (4), 40–50. http://doi.org/10.1016/j. tej.2014.04.002. Ching, L. (2016) Resilience to Climate Change Events: The Paradox of Water (In)Security. Sustainable Cities and Society. Chmutina, K., Lizarralde, G., et al., 2016. Unpacking resilience policy discourse. Cities 58, 70–79. Chou, J.-S., Tu, W.-T., 2011. Failure analysis and risk management of a collapsed large wind turbine tower. Eng. Failure Anal. 18 (1), 295–313. http://doi.org/10.1016/j. engfailanal.2010.09.008. Cohen, S.M., Macknick, J., Averyt, K., Meldrum, J. (2014). Modeling Climate-Water Impacts on Electricity Sector Capacity Expansion Preprint, (May). http://doi.org/ 10.1115/POWER2014-32188. Cradden, L.C., Harrison, G.P., Chick, J.P., 2012. Will climate change impact on wind power development in the UK? Clim. Change 115 (3–4), 837–852. http://doi. org/10.1007/s10584-012-0486-5. Crook, J. a., Jones, L. a., Forster, P.M., Crook, R., 2011. Climate change impacts on future photovoltaic and concentrated solar power energy output. Energy Environ. Sci. 4 (9), 3101. http://doi.org/10.1039/c1ee01495a. ESG International. (1999). EIA Study Haripur Power Project Main Report: Volume 1. Dhaka, Bangladesh. ESG International. (2000). Meghnaghat Power Project 450 MW Combined Cycle Environmental Impact Assessment (Vol. September). Dhaka, Bangladesh. Ebinger, J.O., Vergara, W., 2011a. Climate Impacts on Energy Systems: Key Issues for Energy Sector Adaptation. World Bank, Washington D.C. Ebinger, J., Vergara, W., 2011b. Climate Impacts on Energy Systems. The International Bank for Reconstruction and Development/The World Bank, Washington DC. http://doi.org/10.1596/978-0-8213-8697. Edison Electric Institute (2014). Before and After the Storm—Update: A compilation of recent studies, programs and policies related to storm hardening and resiliency. Retrieved from http://www.eei.org/issuesandpolicy/ electricreliability/mutualassistance/Documents/BeforeandAftertheStorm.pdf. Environmental Resource Management Pte. Ltd Lanco Power International Pte Ltd. (2012). Draft Environmental Impact Assessment BAN: Bhola Gas Power Project. Espinoza, S., Panteli, M., Mancarella, P., Rudnick, H., 2016. Multi-phase assessment and adaptation of power systems resilience to natural hazards. Electric Power Syst. Res. 136, 352–361. http://doi.org/10.1016/j.epsr.2016.03.019. Friend, R., Jarvie, J., et al., 2014. Mainstreaming urban climate resilience into policy and planning; reflections from Asia. Urban Clim. 7, 6–19. Hamilton, M.C., Lambert, J.H., et al., 2016. Resilience analytics with disruption of preferences and lifecycle cost analysis for energy microgrids. Reliab. Eng. Syst. Saf. 150, 11–21. Hammer, S.A., Parshall, L., Leichenko, R., Vancura, P., Panero, M. (2014). Responding to Climate Change in New York State (ClimAID): Chapter 8 Energy. Retrieved from http://www.nyserda.ny.gov/climaid. Hanger, S., Pfenninger, S., Dreyfus, M., Patt, A., 2013. Knowledge and information needs of adaptation policy-makers: a European study. Reg. Environ. Change 13 (1), 91–101. http://doi.org/10.1007/s10113-012-0317-2. Ho, C.-H., Chen, J.-L., et al., 2016. Mitigating uncertainty and enhancing resilience to climate change in the fisheries sector in Taiwan: policy implications for food security. Ocean Coastal Manage. 130, 355–372. Hong Kong Business Environment Council. (2015). Hong Kong Climate Resilience Roadmap for Business Foreword By BEC Chairman. Jaglom, W.S., McFarland, J.R., Colley, M.F., Mack, C.B., Venkatesh, B., Miller, R.L., Kayin, S., 2014. Assessment of projected temperature impacts from climate change on the U.S electric power sector using the Integrated Planning Model1. Energy Policy 73, 524–539. http://doi.org/10.1016/j.enpol.2014.04.032. Japan International Cooperation Agency (JICA) Tokyo Electric Power Services Co. (2015). Preparatory survey on Chittagong Are Coal Fired Power Plant Development Project in Bangladesh Final Report on Power Plant/Port/ Transmission Line/Access Road/Execution Survey of Natural Condition. Japan International Cooperation Agency Tokio Electric Power Services Co Tokyeo Electric Power Company. (2016). Survey on Power System Master Plan 2015 Draft Final Report April 2016. Retrieved from http://powerdivision.portal.gov. bd/sites/default/files/files/powerdivision.portal.gov.bd/page/ 4f81bf4d_1180_4c53_b27c_8fa0eb11e2c1/finalreport_2016.pdf. Jerez, S., Tobin, I., Vautard, R., Montávez, J.P., López-Romero, J.M., Thais, F., Wild, M., 2015. The impact of climate change on photovoltaic power generation in Europe. Nat. Commun. 6, 10014. http://doi.org/10.1038/ncomms10014. Kernaghan, S., da Silva, J., 2014. Initiating and sustaining action: experiences building resilience to climate change in Asian cities. Urban Clim. 7, 47–63.

Kezunovic, M., Dobson, I., Dong, Y., 2008. Impact of Extreme Weather on Power System Blackouts and Forced Outages: New Challenges, In 7th Balkan Power Conf, pp. 1–5. Khan, I., Chowdhury, H., Alam, F., Alam, Q., Afrin, S., 2012. An investigation into the potential impacts of climate change on power generation in Bangladesh. J. Sustainable Energy Environ. 3, 103–110. Khan, I., Alam, F., Alam, Q., 2013. The global climate change and its effect on power generation in Bangladesh. Energy Policy 61, 1460–1470. http://doi.org/10.1016/ j.enpol.2013.05.005. Kiel, J., Petiet, P., et al., 2016. A decision support system for the resilience of critical transport infrastructure to extreme weather events. Trans. Res. Procedia 14, 68– 77. Linnerud, K., Mideksa, T.K., Eskeland, G., 2011. The impact of climate change on nuclear power supply. Energy J. 32 (1), 149–169. Maryono, H., Wijaya, B., et al., 2016. Energy resilience assessment by using SEM approach in the Central Java Province, Indonesia. Procedia Soc. Behav. Sci. 227, 146–151. Meerow, S., Newell, J.P., et al., 2016. Defining urban resilience: a review. Landscape Urban Plann. 147, 38–49. Melillo, Jerry M., Terese (T.C.) Richmond, G.W.Y. (2014). Climate Change Impacts in the United States: The Third National Climate Assessment. Mendiluce, M., 2014. Risky business: Building a resilient power sector. IEEE Power Energy Mag. 12 (5), 34–41. http://doi.org/10.1109/MPE.2014.2331892. Mideksa, T.K., Kallbekken, S., 2010. The impact of climate change on the electricity market: a review. Energy Policy 38 (7), 3579–3585. http://doi.org/10.1016/j. enpol.2010.02.035. Molyneaux, L., Brown, C., et al., 2016. Measuring resilience in energy systems: insights from a range of disciplines. Renewable Sustainable Energy Rev. 59, 1068–1079. Mukheibir, P., 2013. Potential consequences of projected climate change impacts on hydroelectricity generation. Clim. Change 121, 67–78. http://doi.org/10.1007/ s10584-013-0890-5. Mukhopadhyay, S., Hastak, M., 2016. Public utility commissions to foster resilience investment in power grid infrastructure. Procedia Soc. Behav. Sci. 218, 5–12. NTPC Limited (2012). Final Feasibility Report: KHULNA THERMAL POWER PROJECT (2  660 MW) Bangladesh. Dhaka. Retrieved from http://bifpcl.com/new/wpcontent/uploads/2014/06/Feasibility-Report-2013.pdf. Nye, David, 2010. When the Lights Went Out. MIT Press, Cambridge, USA. OECD. (2015). Tools to mainstream adaptation into decision-making processes, in Climate Change Risks and Adaptation: Linking Policy and Economics. Paris. Panteli, M., Mancarella, P., 2015. Influence of extreme weather and climate change on the resilience of power systems: impacts and possible mitigation strategies. Electric Power Syst. Res. 127, 259–270. http://doi.org/10.1016/j. epsr.2015.06.012. Parkinson, S.C., Djilali, N., 2015. Robust response to hydro-climatic change in electricity generation planning. Clim. Change 130 (4), 475–489. http://doi.org/ 10.1007/s10584-015-1359-5. Patt, A., Pfenninger, S., Lilliestam, J., 2013. Vulnerability of solar energy infrastructure and output to climate change. Clim. Change 121 (1), 93–102. http://doi.org/10.1007/s10584-013-0887-0. Rampal power plant deal signed. (2016). Retrieved March 8, 2016, from http:// businessnews24bd.com/rampal-power-plant-deal-signed/. Reuters. (2016). Bangladesh set to pursue coal-power push despite opposition. Retrieved from http://www.dailymail.co.uk/wires/reuters/article-3539721/ Bangladesh-set-pursue-coal-power-push-despite-opposition.html. Rezaei, S.N., Chouinard, L., Langlois, S., Légeron, F., 2016. Analysis of the effect of climate change on the reliability of overhead transmission lines. Sustainable Cities Soc.. http://org/http://dx.doi.org/10.1016/j.scs.2016.01.007. Rubbelke, D., Vogele, S., 2011. Impacts of climate change on European critical infrastructures: the case of the power sector. Environ. Sci. Policy 14 (1), 53–63. http://doi.org/10.1016/j.envsci.2010.10.007. Sailor, D.J., Smith, M., Hart, M., 2008. Climate change implications for wind power resources in the Northwest United States. Renewable Energy 33 (11), 2393– 2406. http://doi.org/10.1016/j.renene.2008.01.007. Schaeffer, R., Szklo, A.S., Pereira de Lucena, A.F., Moreira Cesar Borba, B.S., Pupo Nogueira, L.P., Fleming, F.P., Boulahya, M.S., 2012. Energy sector vulnerability to climate change: a review. Energy 38 (1), 1–12. http://doi.org/10.1016/j. energy.2011.11.056. Senior Correspondent. (2016). Bangladesh all set to build its first LNG terminal in partnership with US firm. Retrieved March 8, 2016, from http://bdnews24.com/ economy/2016/07/18/bangladesh-all-set-to-build-its-first-lng-terminal-inpartnership-with-us-firm. Shahid, S., 2012. Vulnerability of the power sector of Bangladesh to climate change and extreme weather events. Reg. Environ. Change 12 (3), 595–606. http://doi. org/10.1007/s10113-011-0276-z. Singhee, A., Li, Z., Koc, A., Wang, H., Cipriani, J.P., Kim, Y., Gauthier, G.M., 2016. OPRO: precise emergency preparedness for electric utilities. IBM J. Res. Dev. 60 (1), 1–15. The Guardian. (2003, August 12). Heatwave hits French power production. Retrieved from https://www.theguardian.com/world/2003/aug/12/france. nuclear. Tobin, I., Vautard, R., Balog, I., Bréon, F.-M., Jerez, S., Ruti, P.M., Yiou, P., 2015. Assessing climate change impacts on European wind energy from ENSEMBLES high-resolution climate projections. Climatic Change 128 (1-2), 99–112. http:// doi.org/10.1007/s10584-014-1291-0. U.S. Department of Energy (n.d.). Climate Change and the U.S. Energy Sector: Regional Vulnerabilities and Resilience Solutions.

D. Chattopadhyay et al. / The Electricity Journal 29 (2016) 32–41 WBCSD electric utilities. (2014). Building a Resilient Power Sector. Wilbanks, T., Bibello, D., Schmalzer, D., Scott, M. (2013). Climate Change and Energy Supply and Use: Technical report for the US Department of Energy in support of the National Climate Assessment. Washington DC. World Bank, 2014. Turn Down the Heat: Confronting the New Climate Normal. World Bank Publications. van Vliet, M.T.H., Yearsley, J.R., Ludwig, F., Vögele, S., Lettenmaier, D.P., Kabat, P., 2012. Vulnerability of US and European electricity supply to climate change. Nat. Clim. Change 2 (9), 676–681. http://doi.org/10.1038/nclimate1546. Deb Chattopadhyay is an Adjunct Professor of University of Queensland, Australia. He leads the power systems planning group in the World Bank, Washington DC. He is part of a project on climate resilient power systems planning recently initiated in the World Bank.

Evangelia Spyrou is a Ph.D. student at the Johns Hopkins University specializing on treatment of uncertainty in power systems planning. She is developing a stochastic programming model to deal with climate change related uncertainties as part of the World Bank project.

41

Neha Mukhi is a Climate Change Specialist at the World Bank, Washington DC. She is the Task Team Leader of the World Bank project on climate resilience. She holds a Master of Science in Energy and Environment from Harvard University.

Morgan Bazilian is a Visiting Professor at the Royal Institute of Technology of Sweden, and an Associate Researcher at Cambridge University’s Energy Policy Research Group. He is a Lead Energy Specialist at the World Bank.

Adrien Vogt-Schilb is part of the Climate Policy Team at the World Bank. He specializes in the design of effective and acceptable climate mitigation strategies. He has been responsible for developing the robust decision making framework in the World Bank climate resilient planning project. He holds a Master of Energy Economics and a Ph.D. in climate change economics.