How modern systems planning can help tackle demand forecasting risks

By Thu Vu

Summary

• After years of chasing supply gaps, Southeast Asia’s baseload-heavy capacity expansion models have begun to stumble. In some cases, they have become prone to unanticipated implementation risks, resulting in inadequate supply; in others, they have locked power utilities and taxpayers into excess capacity, robbing system operators of the flexibility required to manage lower-than-expected demand at minimal cost to the public.

• In Southeast Asia’s highly dynamic economies, projecting electricity demand growth and required investment in generation has been further complicated by changing national industrial policies, foreign investment agenda, financial market conditions, and energy efficiency and transport electrification trends. This calls for a reevaluation of traditional system planning and power offtake arrangements.

• Planners are now facing more forecasting risk as historical trends break down and there is increased variability around the most sensitive assumptions that typically drive demand models. In response, they may need to embrace strategies to enhance operational and contractual flexibility at the system level to accommodate unpredictable demand patterns. This starts with vigorous testing and a better understanding of the impact of lock-in resulting from different capacity expansion scenarios as well as consideration for dynamic capacity development and power access options.

Rising anxiety

As the summer heat approaches Vietnam’s northern region, there is one burning question in the back of everyone’s mind: how severe will the power cuts be this May and June? Repeated assurances from government leaders and the state utility, Electricity of Vietnam (EVN), have had limited effect in calming nervous residential and industrial customers, particularly the large factories in the electronics manufacturing hubs of Thai Nguyen, Bac Giang and Quang Ninh, that faced rolling blackouts last summer and are worried about the prospects for this year. 

The anxiety is rising for good reason. Electricity consumption is rebounding to double-digit growth – over 16% year-on-year in the first two months of 2024 – as a result of growing industrial output and capital investment in manufacturing by global corporations committed to a “China+1” strategy. Meanwhile, supply continues to remain tight in the region. Limited new capacity is coming onstream and reinforcement grid lines that could otherwise help to bring in additional power from the generators in the south have not yet been completed. In early April, the system operator National Load Dispatch Centre warned that Vietnam’s north could face a supply gap of up to 2.9GW during peak hours this summer, an upward revision from earlier forecasts in January.

While Vietnam’s north is short of generating capacity, other parts of Southeast Asia face the opposite problem. System operators and consumers are struggling with an oversupply of power and high costs for unused capacity. The overbuilt generation capacity has become a financial burden to the national utilities and taxpayers in Indonesia and Malaysia, with the most stress felt in times of depressed demand. 

Since 2010, around 15GW of coal-fired power capacity has been added to Indonesia’s Java-Bali region, which serves 70% of the country’s electricity demand. However, with lower-than-projected demand levels, the main grid registered a 76% reserve margin by 2022, a ratio that the state utility PLN expects will hover between 40-60% in the next decade. Not far away, Peninsular Malaysia’s reserve margin ratio has also surpassed 50%, pushing TNB – Peninsular Malaysia’s only electric utility – to search actively for large new demand sources such as data centres. In both countries, excess generating capacity now stands in the way of national and power sector transition plans, as the incumbent fossil-fuel based generators with long-term offtake and fuel supply agreements are barriers to a nimble pivot toward cost-effective new sources of clean energy. 

These two extreme scenarios are the result of persistent power supply-demand mismatches that highlight the limitations of conventional power system planning and how they leave planners little leeway for handling downside risks. The risks could be in the form of execution delays, or even failure, for pipeline generation projects and unbalanced demand growth (the case of Vietnam). There could also be risks of over-projecting demand growth, resulting in redundant capacity that has to be paid for directly or indirectly by consumers (the case of Indonesia and Malaysia).

Colouring inside the lines

Power system development plans rely on capacity expansion models. Emerging economies in Southeast Asia face high population and industrial activity growth, fast rural electrification and urbanisation rates. This has put constant pressure on national governments to plan for, and secure, more electricity to serve surging demand. 

To respond to these trends, energy planners are tasked with mapping out, in greater detail, how much new capacity will be needed, where the new generators shall be built and how to get these connected to the end-consumers. For years, national power development plans in Southeast Asia have typically focused on detailed lists of power generation and transmission projects slated for development over a 10-, 20-, or even 30-year horizon. Once formalised, the plan becomes the de facto blueprint of the country’s future power system and the primary basis for investments in new generation and transmission assets. 

Furthermore, medium- to long-term power development plans are commonly used by overseas investors to assess future system adequacy and electricity reliability for existing and prospective foreign direct investment projects hosted in the country.  

Long planning horizons are necessary given the lead time for large-scale power plants. A gigawatt-scale coal- or gas-fired power station could take 8-12 years from early development to full commissioning, while the corresponding average figure for an offshore wind farm is nine years. Planning that far ahead, however, comes with inevitable risks.

Any design plans for a future power system are underlined by a set of assumptions and projections. This generally includes, among many other inputs, projections on future demand levels to identify upcoming capacity gaps, a menu of technology options that the model may choose from, data on natural resource potentials relevant to development, and assumed fuel costs and technology-specific capital expenditures to estimate – and optimise – overall system development costs. Any changes to these parameters can influence how the power system under the plan takes shape. 

The significant time lag between the moment the planning is done – including the underlying modelling work – and the 10- or 20-year period that the plan covers, makes it extremely challenging for planners to forecast future market developments with reasonable accuracy. Technological advancements in renewable energy and the restrictive funding environment around fossil fuel assets have derailed many national plans in recent years, while the mass uptake of electric vehicles could potentially lift or reshape the demand profiles of Southeast Asian megacities in the years to come.  

Despite persistent forecast risks, national power development plans have generally been perceived as fixed blueprints that guide project permitting and investment by market regulators and developers. It is also notable that some modelling elements cannot be easily adjusted, making it difficult to introduce ad-hoc changes as market conditions change. For committed fossil-fuel power plants with lengthy development timelines, the investment decisions and offtake agreements are often irreversible once financing has been closed and construction has started – all years before the delivery of the first electrons that might then no longer be needed.

Southeast Asia’s balancing act

One of the most consequential drivers of capacity expansion models is the assumption on the demand growth trajectory. 

Underestimating demand – both in terms of its structure and location – can lead to power shortages, while overestimating demand raises the risk of unused generation capacity. To enable a buffer for possible upside growth, some countries mandate a minimum reserve margin to guarantee system adequacy. For example, Indonesia sets a generous 35% ratio for the Java-Bali grid and a 40% ratio for others, while the Philippines stipulates a minimum 25% for each of its regional grids. 

In the regulated, single-buyer power markets of Indonesia, Malaysia, Thailand, and Vietnam which have favoured large-scale baseload capacity for system expansion, overestimating future electricity demand can result in long-term system-wide rigidities, namely technology lock-in and inflexible payment obligations. This is a consequence of the contractual arrangements that have been used to secure investments in new generation capacity.

Over the past two decades, the state utilities of these countries have relied on independent power producers (IPPs) to expedite the build-out of new, high-capacity power plants. This approach has freed utilities such as PLN, EGAT and EVN from project funding and development responsibilities and risks, and shifted these obligations to private sponsors who commit to bring online new generation assets at a premium. To attract investors, the utilities – as the sole off-taker—typically provide the private sponsors with a 20- to 25-year power purchase agreement (PPA) with explicit take-or-pay clauses that require fixed monthly payments to cover the project’s capital expenditures, fixed maintenance costs, debt repayments, and a reasonable profit margin. Even when no electricity is dispatched, the utilities are still obligated to pay a minimum “capacity charge” to the plant owners to cover for these fixed costs. 

The PPAs and the risk mitigation clauses in project finance loans are structured to shield project sponsors and their lenders from any downside demand and market risks. It was thanks to such protection that IPP gas power plants Ayudhaya and Nong Saeng in Thailand were still ranked as “excellent” in their economic performance by their lenders despite registering an average dispatch ratio of just 28% and 38%, respectively, compared to the initial target of 90%. The projects were originally intended for baseload dispatch but came to serve as peaking plants, a result of lower-than-expected electricity demand growth in Thailand and competition from cheaper hydro power imports.

On the other side of the transaction, these risks are borne fully by the offtakers and taxpayers. In recent years, this risk has materialised in an extreme way for Indonesia’s PLN, which has seen its operating profit suffer as IPPs payments surged from 23% to 34% of operating expenses between 2016-2022. During this period, IPPs bills grew at twice the rate of PLN’s electricity sales revenue.

Can we get it right?

To understand how demand forecasting may need to change, it is important to know how the planners usually arrive at future demand growth estimates.

In Southeast Asia, the primary macroeconomic driver of electricity demand forecasts has been gross domestic product (GDP) growth. The high correlation between economic growth and electricity consumption is supported by broad literature globally. As a result, regional planners have relied on GDP growth targets provided by government macroeconomic planning authorities to make electricity demand growth forecasts. 

Vietnam’s latest power development plan (PDP8) demand figures were based on a targeted annual economic growth rate of 7% between 2021-2030, and a range of 6.5 to 7.5% until 2050, in alignment with the national socio-economic master plan. Indonesia’s electricity master plan (RUKN) and JETP CIPP scenarios assumed a GDP expansion rate of 6% annually to 2030, and nearly 7% in the following decade, pursuant to data from the Ministry of National Development Planning (Bappenas). Meanwhile, in the Philippines’ Power Development Plan 2020-2040, the Department of Energy employed annual GDP projection figures developed by the Development Budget Coordination Committee and the National Economic and Development Authority. These GDP figures are debatable on their own.

Planners then calculate future electricity demand figures by applying a multiplier that is believed to capture the elasticity of electricity demand to economic growth. Some countries use more granular regional historical trends to set the multiplier – an approach taken by the Philippines which applied a constant 0.8 multiplier to the Luzon grid, and a 1.0 ratio for Visayas and Mindanao grids, to forecast power demand growth until 2040. 

By contrast, planners in Indonesia and Vietnam assume that the multiplier will gradually trend down as income levels grow past a certain threshold, a pattern that has been observed in many markets. Vietnam’s electricity consumption-to-GDP growth ratio averaged 1.6 between 2010-2019, but national planners expect this to come down to 1.2 in the latter part of this decade, and further below 1.0 between 2031-2040. In contrast, in Indonesia, counterparts believe that the elasticity will peak by 2030 to 1.3, from the current average of 1.13, due to increased industrial activity and electrification in industry and transport, before gradually dropping to around 1.0 by 2040 and further after.  

Despite its importance in the planning process, deciding on the exact elasticity multiplier is not a straightforward task. Historical data suggests that the correlation between electricity consumption and economic growth can be highly volatile and with little predictive power. 

To complicate matters further, an emerging set of demand drivers is introducing a mix of push and pull factors that could impact how national electricity demand curves take shape in the future. Some noteworthy trends that have been observed in Southeast Asia in recent years are based on structural changes in power system technology and design. While the pace of uptake is hard to predict, the upward potential for adoption is high. These changes are unfolding as Southeast Asian power systems become central to low-carbon transition aspirations; they are now shouldering new responsibilities and can trigger far-reaching policy responses and behavioural changes. 

• Distributed solar photovoltaic (PV) systems have already proven to be quickly scalable in markets such as Thailand, Malaysia, and Vietnam, driven by policy support, the improving economics of solar PV, and the growing appetite for clean electricity access by commercial and industrial consumers. Despite the progress to date, rooftop solar PV remains a vastly untapped “low-hanging fruit” that has the potential to attract incremental and mass-scale investments once regulatory hurdles are removed. Across the region, the main stumbling block has been on the policy front. But the good news is that regulators and power utilities have now acknowledged that distributed solar PV systems will become an inevitable part of national power systems and have shown increasing willingness to work with the industry to simplify permitting processes and stimulate further deployment. At scale, distributed solar PV systems can distort on-grid electricity demand profiles, and will require system-level planning and technical adjustments from system operators. 

• Electric vehicles (EVs) adoption, from two-wheelers to passenger and commercial vehicles, is picking up pace in Southeast Asia, supporting the electrification of road transportation and national net-zero emissions strategies. Sales figures indicate that 2023 was a pivotal year for EV penetration in the region, with exponential growth registered in key markets. At over 100,000 units, Thailand’s new EVs registrations were 3.8x the previous year. In Malaysia, sales of battery and hybrid vehicles jumped 69% year-on-year, to 38,000 units. The Philippines also saw a ten-fold increase in new EVs sales, at over 10,000 units. Imported cars, particularly of Chinese brands, have turbocharged this growth, but homegrown manufacturers such as Energy Absolute (Thailand) and VinFast (Vietnam) have been key in supporting the development of EVs infrastructure and enabling policies in their respective countries. Several regional governments have released EV penetration targets and tax incentives. Energy ministries will need to catch up. This shall include not only upward revisions to demand forecasts, but also designing of charging tariff schemes for system-wide benefits, such as reducing peak load and variable renewable energy curtailment.

• Data centres have emerged as a new digital infrastructure investment area that Southeast Asian governments are keen to promote. Broad digitalization and a uniform push for data localization rules across the region are spurring demand for onshore data centres. Industry analysts have projected double-digit annual growth potential for the nascent data centre markets of Indonesia, Malaysia, and Vietnam. These discussions, however, have not been in sync with power planning processes, despite the significant carbon footprint and heavy energy consumption profile of data centres. With data localisation remaining firm on the agenda of ICT ministries, energy planners will inevitably need to make room for this new group of consumers. A point of reference is Singapore, which has balanced its data centres hosting needs with power stress concerns by imposing stringent energy efficiency and sustainability criteria on new facilities.   

Regional planners can exert high caution when accounting for new variables and trends in electricity demand growth projection. That’s why it’s so important for investors and analysts to understand the constraints that are often imposed during the modelling stage. These implicit planning decisions shape the usefulness of a PDP and its credibility to match or respond to actual developments.

Planning for flexibility

Faced with increasing forecasting risks, power system planners in Southeast Asia may need to embrace strategies to enhance operational and contractual flexibility at the system level to accommodate unpredictable demand patterns. 

From a modelling perspective, demand sensitivity analysis is possible in capacity expansion models. A robust examination of system development pathways at different demand growth levels can shed light on potential technology lock-in risks, and provide guidance on mitigation or response strategies under certain demand scenarios. TransitionZero’s Model Builder will enable this type of stress testing for system development scenarios by giving users the ability to set demand growth assumptions at the values of choice. 

A second target application for Model Builder is grid design and dispatch. One of the reasons why markets like Vietnam, Indonesia or Malaysia have been unable to balance demand-supply gaps is limited grid connectivity with nearby regions or power markets that could have otherwise facilitated complementary resources and seasonal or time-of-day demand differences. Socio-economic-political resistance towards high-value investments in long-distance transmission and towards deeper cross-border power exchange have dominated the regional power landscape for decades, but member states might have realised by now the hefty price tag associated with the status quo. The Model Builder aims to support a better understanding of this trade-off by enabling users to compare different system development scenarios under varying transmission capacity options between the markets.