System Adequacy Beyond Megawatts: adequacy is a bundle of grid services, not just MW totals

Abstract

Under IRP 2025, South Africa’s coal fleet is scheduled for retirement on a 50-year station-life basis, creating a major decommissioning cliff between 2029 and 2035 against a 1.4% annual demand growth projected in the MTSAO 2026–2030 base case. This paper argues that coal decommissioning must not proceed until three distinct prerequisites are met:

• Transmission: Grid capacity must be fully commissioned, not merely planned, to evacuate and deliver new generation to demand centres
• Procurement: Replacement capacity must achieve financial close, legal contract, and clear technical specification. It must explicitly replace the specific ancillary services lost, rather than being treated as an undifferentiated MW total
• Execution: Committed capacity must demonstrate high-confidence commercial operation dates (CODs) verified via milestone tracking, backed by a strategic-reserve overlap period

These prerequisites are critical because the choice to decommission a coal unit is materially asymmetric to a deferral decision. Furthermore, current Minimum Emission Standards (MES) exemptions are legally anchored to these specific closure dates. Retiring coal without meeting these conditions shifts manageable planning risks into an irreversible, structural security-of-supply deficit.

System adequacy is a bundle of services, not a single MW total

Coal retirement is frequently modelled as a single-dimension capacity balance: MW retired against MW added. This understates the replacement task. The South African Grid Code’s Ancillary Services Technical Requirements define system adequacy as a bundle of distinct services: resource adequacy (installed MW and MWh), flexibility (ramping, fast start, cycling), frequency control (instantaneous reserve activated within 10 seconds; regulating reserve across the 49.85–50.15 Hz band), voltage and reactive-power support, fault current and system strength, and restoration capability including black start.

The System Operator must hold frequency above 49.5 Hz following any single credible contingency, the largest of which is defined as the loss of a full-load Koeberg unit (920 MW). The contingency standard itself is sized on the loss of one very large synchronous machine, meaning the inertia, fault level, and voltage support it provides are concentrated in a single location. Technologies that differ on these dimensions cannot be aggregated into an undifferentiated "replacement capacity" figure without losing information the System Operator treats as decisive.

Nuclear and Gas-to-Power replace opposite parts of the coal adequacy bundle

When replacing retiring coal units, new nuclear and GTP assets function as complementary opposites rather than direct substitutes for the same megawatts:

• Nuclear Baseload: Replaces coal’s high-capacity factor energy, heavy synchronous inertia, and localized system strength. However, it lacks operational flexibility, cannot fast-start, and requires dedicated grid corridors for startup since it cannot initiate a black start (despite its ability to island to house load during a collapse)
• GTP: Replaces coal's mid-merit flexibility, rapid ramping, and fast-start reserve capabilities. Despite fuel price volatility, flexible gas infrastructure offers strong system economics by displacing expensive diesel and providing the dispatchable backing needed to scale variable renewables on a coal-heavy grid.

Prerequisite (ii), firm replacement-capacity commitment must therefore be assessed service by service and, where a capability is location-bound, location by location. A contracted MW of flexible GTP is not evidence that a retiring coal unit's inertia and voltage support have been replaced, just as a MW of rigid nuclear base load cannot substitute for the dynamic flexibility needed to stabilise a changing grid.

The adequacy gap is quantified, not hypothetical

The adequacy gap is a quantifiable structural reality. Author modeling shows that under a strict 50-year retirement schedule with new gas assumptions excluded, South Africa faces a structural adequacy gap of 19 TWh by 2030, widening to 35 TWh by 2040. The root cause is a firm-capacity replacement-timing problem—where dispatchable capacity retires faster than it is built—rather than a failure of renewable deployment. Even a 60-year "Delayed Shutdown" pathway, which retains an incremental 1.8 GW of coal by 2038 and 3.5 GW by 2040 (operating at a 60% load factor), only partially offsets the deficit. This life-extension strategy closes 8.6 TWh in 2038 and 18.5 TWh in 2040, leaving heavy residual gaps of roughly 25 TWh and 17 TWh respectively. Consequently, coal life extension functions strictly as a short-term bridge; it narrows but cannot close the deficit, leaving new firm capacity—principally Gas-to-Power—to secure the remaining grid balance.

Prerequisite 1: Transmission readiness: commissioned capacity, not planned capacity

New generation cannot clear grid bottlenecks without commissioned transmission infrastructure, regardless of its contractual status. The NTCSA's Transmission Development Plan (TDP 2025–2034) requires 14,500 km of new high-voltage lines and over 200 transformers to integrate 56 GW of new power by 2034, costing an estimated R440 billion (with R112 billion required in the first five years). Delivery is severely behind schedule: achieving this target requires a peak build rate of 1,400 to 2,300 km of line per year, vastly eclipsing the historical average of 400 km; in 2024/25, Eskom missed its modest 89 km target. Global equipment demand further escalates capital costs and execution risks. This grid constraint is already active, leaving over 3,000 MW of committed renewable capacity stranded. Therefore, the criteria for this prerequisite must be actual commissioning status—lines and transformers must be energized and available for dispatch, not merely listed as a planning assumption in the TDP.

Prerequisite 2: Firm replacement-capacity commitment: a bankable obligation, differentiated by service

This prerequisite verifies whether replacement capacity exists as a committed, legally binding obligation—financially closed, contracted, and fully permitted with secured fuel supply—and matches the specific grid service (baseload, flexibility, or inertia) being retired. The proposed 2,500 MW new nuclear build demonstrates how theoretical capacity fails this test: its Section 34 determination was withdrawn in August 2024, feasibility studies remained incomplete into 2025, and the Request for Proposals (RFP) has faced recurring bureaucratic delays. Similarly, gas-to-power planning illustrates the danger of treating unclosed bids as firm assumptions. A prime example is the Risk Mitigation IPP programme, which culminated in the High Court formally cancelling Karpowership's generation licenses in July 2025 following years of environmental rejections and litigation. Unclosed, unpermitted projects cannot be considered firm replacement assets. Furthermore, capacity tracking must be executed service by service and location; generic MW aggregation creates a dangerous planning illusion that risks systemic supply failure.

Prerequisite 3: Schedule and delivery confidence: verified alignment of operational dates

This prerequisite applies only after capacity clears the commitment test, verifying whether replacement assets will realistically achieve commercial operation (COD) on the dates assumed by adequacy models. Historically, South African timelines for both candidate technologies are systematically over-optimistic. Initial planning placed first new nuclear commissioning in 2032/33, yet current frameworks shift a 2,500 MW block into 2031–2035, followed by 1,925 MW toward 2040, dates that remain highly volatile given outstanding RFPs. Similarly, GTP capacity legally bound to commission within 12 months of financial close has been routinely deferred by years due to environmental and permitting litigation. Because a single schedule slippage can trigger a severe firm capacity deficit, this constraint carries the highest consequence. The choice to decommission is physically asymmetrical: a coal unit held in operational reserve pending schedule confirmation can be safely retired once replacement power arrives, whereas an idled unit cannot easily be recovered. Once a thermal asset transitions beyond warm shutdown into cold lay-up or full decommissioning, boiler, turbine, and rotor degradation quickly becomes irreversible.

Regulatory sequencing: MES exemptions are fixed to arbitrary dates, ignoring replacement readiness

The environmental compliance regime exposes a critical scheduling conflict. In March 2025, the DFFE Minister granted Eskom limited Minimum Emission Standards (MES) exemptions for eight coal stations, tying them strictly to calendar deadlines rather than system adequacy. Duvha and Matla received exemptions running to their 2034 retirement targets, while Kendal, Lethabo, Majuba, Matimba, Medupi, and Tutuka face hard limits expiring on 1 April 2030. These exemptions carry binding conditions, including mandatory emissions-dispatch pricing and flue-gas desulphurisation reviews. The regulatory logic presumes these stations will shut down on schedule; however, NEMA:AQA lacks a mechanism to defer decommissioning if transmission infrastructure, contracted capacity, or schedule confidence fail to materialise by 2030. If these prerequisites are unmet, Eskom faces an impossible choice: shut down vital units to comply with the law, or operate in breach of an exemption framework already facing active High Court challenges from civil society. To secure the grid, energy policy must urgently re-link MES timelines to verified milestone readiness rather than rigid calendar dates.

A Readiness-Based Approach to Coal Retirement

The choice confronting policymakers is not whether to transition away from coal, but whether to execute that transition through planned milestones or forced grid failures. The current calendar-driven schedule is mathematically and operationally unviable. To resolve this impasse, the Department of Mineral Resources and Energy and the DFFE must merge their independent planning tracks into a single, unified regulatory mechanism. Coal retirement dates and MES exemption expiries must be legally decoupled from fixed calendar dates and explicitly bound to a three-test readiness gate: energized transmission lines, closed service-specific contracts, and verified commercial operation dates. Transforming these prerequisites into statutory triggers protects the country from irreversible energy deficits. It forces the state to prove that a stable replacement system is actively running before the old one is legally and physically dismantled.