Choosing a reactor technology, systematically

What the RTA method is

RTA is a decision-making method. It does not tell you which reactor is "best" in general; it tells you which of the available designs best meets your stated objectives, and it leaves a written trail showing why. IAEA §1.2, §2.2

The process in seven steps

This sequence combines the IAEA pre-RTA steps and matrix with EPRI's screening funnel. Do the steps in order. EPRI warns that objectives not fully worked out at the start can taint the results of the whole assessment, and the IAEA requires the rationale for each weight to be written before the weights are assigned. EPRI §2.2 IAEA §5

1. Fix the objectives

   Write down, and have decision makers approve, what the programme or project must achieve — usually four to eight objectives — and rank them. IAEA §3.1–3.2 EPRI §2.2

2. Turn objectives into criteria

   Map each objective to the KEs and KTs that measure it. Mark which criteria are hard requirements (knock-outs) and which are preferences to be scored. IAEA §3.1 EPRI §2.1, §2.5

3. Build the long list and screen it

   Survey available technologies, remove those that clearly cannot meet the requirements (with a written reason), and narrow to three to five comparable candidate designs. IAEA §3.3, §4.2 EPRI Steps 2–5

4. Set the weights

   Write a rationale for each KE's importance, then assign KE weights (sum 100%) and KT weights within each KE (sum 100%). Do this before looking at vendor data. IAEA Annex, Fig. A-1

5. Define the scoring scales

   For every KT, write what earns a 1, 2, 3, 4 and 5, using one scale throughout and a realistic range of values. IAEA §4.1, Table 6 EPRI §2.4

6. Collect evidence and score

   Send every vendor the same questions, score each design against the rubric, and record the justification for every score. IAEA §4.3, Annex

7. Aggregate, test, decide, document

   Compute totals, test sensitivity to the weights, compare the issues numbers cannot capture, check the result against the original objectives, and write it up. IAEA §2.2.3, Table 5 EPRI §2.1.6, §2.3

The funnel

Effort increases as the field narrows: cheap, public-information screening first; detailed vendor engagement only for the few designs that survive. EPRI §2.1

1. Technologies of interestPWR, BWR, PHWR, HTGR, SFR, LFR, MSR, SMR types … — broad, from public data
2. Candidate technologiesKnock-out and avoidance screening
3. Potential designsVendor-specific designs of those technologies
4. Candidate designs · 3 to 5Second screening; engage vendors, NDAs
5. RTA matrixWeighted scoring on KEs / KTs
6. Proposed design + alternativesRanked, stress-tested, documented

When RTA is used

RTA is not a one-off study. It runs alongside the IAEA Milestones approach and is repeated with more detail at each phase; issues found at one phase are carried into the next. IAEA §1.1, §2.1

Which key elements are included in each phase

From IAEA Table 7. Phase 1 usually omits KE1 (no site chosen yet) and KE7 (safeguards assumed to be met). By Phase 3, readiness and safeguardability have been settled with the vendors or the design eliminated. Economics is often hard to quantify early or handled by a separate team. KE2–KE5 can always differentiate technologies and belong in every phase.

Before you score

The IAEA calls these the pre-RTA steps. They decide the quality of everything that follows. IAEA §3

4.1 Write the policy objectives

Start from the national energy policy (or, for a utility, the business case) and answer three questions in writing:

• Under what conditions does this programme or project make sense?
• Why was this course of action chosen?
• Which specific outcomes will count as success?

Then select the objectives that matter most — typically four to eight — rank them, and have the policy maker assign relative weights that sum to 100%. Typical candidates include: maintain nuclear safety performance; produce electricity at competitive cost; meet the energy plan's capacity and timing; build national participation and technology transfer; use proven technology; secure long-term vendor support, components and fuel; develop local industry and people; minimise environmental impact; and keep schedule and financing cost under control. IAEA §3.1–3.2, Tables 3–4

For an owner-operator, EPRI frames the same step as mission and business objectives: end-product (electricity, heat, hydrogen …) and its business case, required output, target location, service requirements (firm or flexible, black start, resilience), operating life, need dates, target budgets, and long-term goals such as fleet build-out. EPRI §2.2

4.2 Derive the criteria

Translate each objective into technology features that can be measured, using established requirement sets as a menu rather than a checklist: the IAEA Common User Considerations (NP-T-2.1), the European Utility Requirements, and EPRI's Utility Requirements Documents for advanced LWRs and SMRs. Do not forget non-technical factors — vendor construction record, technology transfer history, political considerations, national resources — which can dominate a decision. IAEA §3.1, Tables 1–2

4.3 Classify every criterion

EPRI's three-tier classification is the cleanest way to keep hard requirements out of the scoring. Word them precisely: do not write a preference as a requirement, or the reverse. EPRI §2.1, §2.5

4.4 Build and screen the long list

• Sources. The IAEA Advanced Reactors Information System (ARIS), licensing submissions and environmental assessments published by regulators, third-party studies, and vendor information. In Phase 2, detailed data normally requires an NDA and a uniform questionnaire sent to every vendor. IAEA §4.3
• Eliminate first, with reasons. Remove unsuitable technologies with a written justification tied to the objectives before any scoring. IAEA §4.2
• Compare like with like. One large unit and several co-located SMRs can deliver the same ~1200 MWe but behave very differently on many KTs. Run separate RTAs for large WCRs and for SMRs, and decide "large or small" at a higher level. IAEA §4.2
• Aim for three to five candidates. If only one design survives, do not treat it as the answer: relax avoidance factors or carry the best rejected design forward as a comparator. EPRI §2.1.5
• Compare the same scope. Some vendors offer only the nuclear island. Make sure every design is assessed over the same plant boundary. EPRI §2.2

4.5 Set up the team and the paperwork

• The NEPIO (Phase 1) and the owner organization (Phase 2) take full responsibility for the RTA; the team reports to the top decision makers. IAEA §2.2.2
• Collectively the team needs full expertise in design, engineering, construction and operation, plus working knowledge of fuel, fuel cycles and waste. EPRI adds roles for transmission planning, communications and commercial contracting, alongside an executive manager, project manager, technical specialists and existing plant operating staff. IAEA §2.2.2 EPRI App. B
• Consultants and universities fill gaps, but their reports are inputs to the team, not the sole basis for a decision. IAEA §2.2.2
• Quality rests on three things: well-defined national requirements, enough objective information on each design, and a team qualified across all KEs. Set document control and review/approval rules at the start. IAEA §2.2.3
• Plan for the effort: EPRI estimates a full owner-operator assessment takes six months to two years. EPRI Exec. Summary

Running the RTA matrix

The RTA matrix is a table with one row per key topic, a weight for each, and a score column per design. The IAEA Annex sets out the order of work below; the workbench implements it. IAEA Annex, Fig. A-1

1. Weight the key elements

   Write a short rationale for each KE's importance to your objectives, then assign percentages summing to 100%. IAEA Table 7 gives High / Medium / Low starting points for large WCRs, SMRs and non-electric use (shown in section 7). A "Low" rating reflects how little a KE is expected to separate the designs in that setting — not that the subject is unimportant.

2. Weight the key topics inside each element

   Write a rationale per KT and assign weights that sum to 100% within the KE. Split a KT into sub-topics if needed (the Annex divides KT 1.2 into wind and seasonal variation, 50% each). Give a KT zero weight when it does not apply — KT 8.1 (SMR readiness) for large WCRs, for example. Weights may also reflect how much information is realistically available.

3. Write the scoring rubric

   For each KT, define what earns each score from 1 to 5, anchored to your site and requirements (the Annex, for a site needing a 0.2 g design basis, uses "5: seismic design 0.3 g … 1: cannot withstand 0.12 g or no information"). Two rules keep the matrix unbiased:

   • One scale everywhere. Never give important topics a wider range (1–10) and minor ones a narrower range (1–5); importance belongs in the weights. IAEA §4.1
   • A realistic range. Anchor scores to the range real designs occupy — see the capacity-factor example in section 6. When the data cannot separate two designs with confidence, score them equally. IAEA §4.1, Table 6

4. Collect evidence and score

   Send every vendor the same questions (the key element reference lists examples). Score each design against the rubric and write one line of justification per score. Mark scores that await vendor answers as provisional (in the workbench, use the notes field). A score of 1 for "no information" is given only after the vendor has been asked and has not answered adequately. IAEA Annex, note to Table A-1

5. Aggregate

   Multiply each score by its KT weight and sum within the KE; multiply each KE result by the KE weight and sum across KEs. The result is a total on the 1–5 scale for each design. IAEA §4.1 EPRI §2.3

6. Stress-test

   Re-run the totals with different weight sets — for example the weights preferred by different stakeholder groups — and check whether the leader changes. Then compare the qualitative issues the numbers cannot capture, and check each finalist against every original objective (exceeds / meets / partially meets / does not meet). EPRI §2.1.6, §2.3, Table 2-4

7. Decide and document

   Name a proposed design and at least one alternative. Record the weights and their rationale, the rubrics, the evidence behind each score, the sensitivity results and the open questions. This record carries into the feasibility study, bid specification and bid evaluation. IAEA §2.2.3 EPRI §2.1.6

Build your matrix in the workbench

One key element, scored end to end

The IAEA Annex scores three large WCR designs for the fictitious country "Retasland" on KE1, Site and environment. The weights and scores below are from Table A-1; the weighted sums are calculated here.

* The Annex marks these scores "more info" — provisional until the vendor answers. KT 1.2 combines two sub-topics weighted 50/50 that received identical scores.

How to read it:

• Arithmetic. NPP1 = 0.25×5 + 0.25×5 + 0.05×4 + 0.05×5 + 0.05×5 + 0.15×1 + 0.05×5 + 0.15×4 = 4.20.
• Provisional scores matter. NPP2 trails mainly on the two heaviest topics — and both of those scores are waiting on vendor information. Its KE1 result could move a long way once the questions are answered.
• Watch for hidden knock-outs. NPP1 leads despite scoring 1 on site size (15% weight). If its footprint truly cannot fit the site, that is an exclusionary issue, not something a weighted average should absorb.
• Roll-up. KE1 is one element. If KE1 carried, say, 15% of the total, these results would contribute 0.63, 0.38 and 0.45 points to the three totals.
• Justify every score. The Annex records reasons such as a heaviest component of 330 t × 3.85 m high (NPP1, scored 5) versus 700 t × 5 m (NPP3, scored 3) for transport over the site's roads and bridges.

Setting an honest scale: capacity factor

Three designs have capacity factors of 75%, 88% and 94%. Scored on a 0–100% scale in 25-point steps, all three crowd into the top two scores and the scale barely separates them. Scored on a relevant range (below 80% = 1, 80–84% = 2, 85–89% = 3, 90–95% = 4, above 95% = 5) they score 1, 3 and 4 — the difference that matters becomes visible. IAEA §4.1, Table 6

The 10 key elements

What each element covers, when it separates designs, typical importance, its key topics and example questions for vendors. Topic descriptions are condensed from IAEA §5; open an element to see the detail.

Expand all

Importance ratings are IAEA's examples, compiled from Member State feedback and expert opinion. The IAEA is explicit that the policy team sets the actual ranges, values and rationale.

Fitting the weights to your situation

Your circumstances, not the reactor catalogue, decide which elements carry weight. These are the most common shifts, each grounded in the source documents.

Reading the result

The ranked totals are a guide, not the decision. EPRI is explicit that the proposed design should reflect everything known about the candidates, not the numerical score alone. EPRI §2.1.6 Before naming a winner, run these five checks.

1. Completeness

   Every design is scored on every weighted topic, with evidence. Treat any total that rests on missing or provisional scores as provisional.

2. Knock-outs

   Look for low scores (1–2) on heavily weighted topics. A good average can hide a disqualifying weakness; if it is disqualifying, handle it as an exclusion. EPRI §2.4

3. Margin versus uncertainty

   If the gap between the top two is small compared with how uncertain the scores are, treat it as a tie and let the qualitative comparison decide. Claims for designs without operating history carry more uncertainty than records from operating fleets. IAEA KT 3.7, KT 4.2

4. Sensitivity

   Change each KE weight up and down, and try the weight sets of different stakeholders. A leader that survives is a robust choice; one that flips needs a sharper argument. EPRI §2.3

5. Issues and alignment

   Tabulate issues that do not reduce to a number (fuel qualification, site fit, remaining design work, grid flexibility) side by side, then rate each finalist against every original objective. EPRI Tables 2-3, 2-4

Common pitfalls

IAEA and EPRI side by side

The two documents share one logic — objectives, screening, weighted scoring, judgement — at different levels. Use the IAEA structure for the matrix; borrow EPRI's screening vocabulary and decision checks.

Where EPRI's criteria sit in the IAEA matrix

Your RTA checklist

Tick items as you go; progress is saved in this browser only.

0 done · Reset

Prepare

• Policy or business objectives written, ranked, weighted and approved by decision makers.
• Criteria derived from the objectives; each marked exclusionary, avoidance or suitability.
• RTA team formed across all key elements, reporting to top management; consultants' roles defined.
• Document control, review and approval rules agreed.

Screen

• Long list built from ARIS, regulator submissions, studies and vendors.
• Unsuitable technologies removed, each with a written reason.
• Separate assessments for large reactors and SMRs where both remain.
• Three to five comparable candidate designs over the same plant scope.

Score

• KE and KT weights set, with rationale, before any vendor data is scored.
• A 1–5 rubric written for every KT, on one scale and a realistic range.
• Uniform questionnaire sent to every vendor (under NDA where needed).
• Every score justified; provisional scores flagged and followed up.

Decide

• Totals computed; low scores on heavy topics reviewed as possible knock-outs.
• Sensitivity tested with alternative weight sets.
• Issue comparison and alignment with each objective completed.
• Proposed design and alternatives documented, with open questions for the bid stage.

Glossary

RTA

Reactor technology assessment — the IAEA decision-making methodology for selecting the most suitable reactor technology.

KE / KT / ST

Key element / key topic / sub-topic — the three levels of the RTA criteria hierarchy.

SMART

Simple multi-attribute rating technique — weight the criteria, score the options, add up. Alternatives include AHP, TOPSIS and ELECTRE.

NEPIO

Nuclear energy programme implementing organization — leads Phase 1 in a newcomer country.

PFS / FS

Pre-feasibility study (Phase 1) / feasibility study (Phase 2).

BIS

Bid invitation specification.

WCR

Water-cooled reactor; "large WCR" means large, grid-scale water-cooled units (PWR, BWR, PHWR), as distinct from SMRs.

SMR

Small modular reactor; definitions vary, commonly up to about 300 MWe per unit. Microreactors are commonly defined as up to about 50 MWe.

FOAK / NOAK

First-of-a-kind / Nth-of-a-kind plant.

NSSS / BOP

Nuclear steam supply system / balance of plant (turbine island and support systems).

DiD

Defence in depth — five levels of protection defined in IAEA SSR-2/1 (Rev. 1), kept as independent of each other as practicable.

DSA / PSA

Deterministic safety analysis / probabilistic safety assessment.

CDF / LERF

Core damage frequency / large early release frequency — typical PSA safety goals.

SAR / OLCs

Safety analysis report / operational limits and conditions (technical specifications).

O&EPs / SAMGs

Operating and emergency procedures / severe accident management guidelines.

HALEU

High-assay low-enriched uranium, enriched between 5% and 20% U-235.

MOX

Mixed uranium–plutonium oxide fuel.

EPZ

Emergency planning zone.

UHS

Ultimate heat sink.

TRL

Technology readiness level, a maturity scale used by EPRI and others.

OCC / LCOE

Overnight capital cost (excluding financing) / levelized cost of electricity (or energy).

PPE / SPE

Plant parameter envelope (bounding values across several designs) / site parameter envelope (one design's site requirements).

ARIS / PRIS

IAEA Advanced Reactors Information System / Power Reactor Information System.

CUC / EUR / URD

IAEA Common User Considerations / European Utility Requirements / EPRI Utility Requirements Document.

Sources and further reading

Primary sources

• International Atomic Energy Agency (2022). Nuclear Reactor Technology Assessment for Near Term Deployment. IAEA Nuclear Energy Series No. NR-T-1.10 (Rev. 1). Vienna: IAEA.
• Electric Power Research Institute (2022). Advanced Nuclear Technology: Owner-Operator Reactor Technology Assessment Guide — 2022 Version. Report 3002025344. Palo Alto, CA: EPRI.

Referenced by those sources

• IAEA (2015). Milestones in the Development of a National Infrastructure for Nuclear Power. NG-G-3.1 (Rev. 1).
• IAEA (2011). Invitation and Evaluation of Bids for Nuclear Power Plants. NG-T-3.9.
• IAEA (2009). Common User Considerations (CUC) by Developing Countries for Future Nuclear Energy Systems. NP-T-2.1.
• IAEA Advanced Reactors Information System (ARIS): aris.iaea.org · Power Reactor Information System (PRIS): pris.iaea.org
• EPRI (2022). Site Selection and Evaluation Criteria for New Nuclear Energy Generation Facilities (Siting Guide) — 2022 Revision. 3002023910.