Hydrogen-Fired Furnaces: How to Qualify Refractories Before the Fuel Switch

The Direct Answer: Do Not Buy a “Hydrogen-Ready” Grade Before Defining the New Exposure

A furnace does not become ready for hydrogen because its burners can ignite a hydrogen blend. The fuel conversion changes a connected system: flame speed, burner momentum, gas volume, heat-release position, radiative and convective transfer, oxygen potential, water-vapor concentration, product oxidation, scale formation, seal leakage, cold-zone condensation and control philosophy. The refractory lining experiences the combined result, not the hydrogen percentage written on the project schedule.

The correct starting point for hydrogen furnace refractories is therefore an exposure definition. The plant must identify what the hot face, joints, anchors, insulation and shell will actually experience in each zone during steady operation, turndown, ignition, trip, purge and shutdown. Only then can a laboratory program or supplier proposal demonstrate that a lining is suitable.

This distinction prevents a common procurement error. A product described as resistant to hydrogen may have been tested in dry, reducing H₂/inert gas, while the intended application is complete hydrogen combustion with a high partial pressure of steam and excess oxygen. Another product may survive a uniform laboratory atmosphere but fail beside a burner because the converted flame creates a local heat-flux peak. The words “hydrogen service” can describe fundamentally different chemical and thermal environments.

A reliable hydrogen-fired furnace lining is not selected from one property table. It is qualified through a chain of evidence: a zone-based exposure map, combustion and heat-transfer scenarios, reaction hypotheses, representative laboratory tests, instrumented field specimens, controlled fuel steps, inspection results and clearly defined operating limits.

First Separate Two Hydrogen Questions That Are Often Blended Together

Hydrogen as a Fuel

When hydrogen is burned to provide industrial heat, complete combustion produces water vapor rather than carbon dioxide at the point of combustion. The relevant lining environment may therefore contain more steam, different residual oxygen and a different temperature and heat-flux field from the natural-gas baseline. Air-fired and oxyfuel cases are not equivalent because nitrogen dilution, flue-gas volume, emissivity, recirculation and burner design can differ substantially.

The engineering question is not simply whether a refractory is stable in hydrogen gas. It is whether the entire lining system remains stable in the actual hydrogen combustion atmosphere, including the water vapor generated, the oxygen not consumed, any local pockets of unburned hydrogen and the solids or vapors released by the heated product.

Hydrogen as a Reducing Agent or Process Gas

Hydrogen-based direct reduction, shaft furnaces and emerging smelting routes present a different problem. In those systems, the gas may intentionally remain strongly reducing and may interact with ore, partially reduced solids, dust, carbon monoxide, carbon, metal and slag. Gas penetration and reduction of refractory constituents can become more important than the steam-rich products of complete combustion.

A review of refractory challenges for hydrogen implementation in iron and steelmaking emphasizes both reducing hydrogen exposure and the water vapor created when hydrogen is used. These are related but not interchangeable cases. A refractory reducing atmosphere specification must therefore state gas composition, pressure, temperature, flow, moisture and process-material contact rather than use the generic label “H₂ resistant.”

Why the Distinction Changes the Material Shortlist

A dense high-alumina material with a silicate-rich bond may respond differently from a low-cement corundum castable, a magnesia-spinel brick, a silica lining, a phosphate-bonded repair material or a carbon-containing basic refractory. Potentially reactive minor phases can be more important than the headline alumina or magnesia percentage. Bond chemistry, impurity distribution, open porosity, permeability, carbon content, metal additions, joint design and temperature gradient must be evaluated together.

This is consistent with the broader refractory material selection service-envelope method: chemistry is only one part of the decision. Geometry, restraint, installation, cycling, load, repair practice and operating variability determine whether a chemically acceptable material becomes a reliable lining.

The Fuel Switch Alters Six Variables at the Same Time

1. Water-Vapor Loading

Hydrogen combustion can materially increase the water-vapor fraction of furnace gas. The consequence is not a universal “steam attack” mechanism; it is a change in chemical potential and transport conditions. Steam can participate in oxidation, hydroxylation, volatilization, hydration or scale-growth reactions depending on material, temperature and local gas composition. The meaningful question is which phase reacts, at what temperature, at what rate, and whether the products remain protective or create porosity and strength loss.

This is why water vapor refractory corrosion cannot be assessed with a single room-temperature moisture test. Hot-face reactions, intermediate-temperature exposure and cold-face condensation occupy different regimes. A material stable at 1,200°C may still develop risk during prolonged startup at 300–600°C, while an insulation layer protected during normal service may encounter condensate after a trip or during seasonal shutdown.

2. Local Oxygen Potential

Average stack oxygen does not describe the atmosphere beside every burner or behind every load. Fuel staging, mixing quality, burner turndown, air leakage and transient control can create oxidizing, neutral and reducing pockets in the same furnace. Hydrogen's combustion characteristics can make flame stabilization and mixing strategy especially important during conversion.

Local oxygen potential influences iron-bearing impurities, carbon-containing refractories, metal fibers, anchors, protective oxide layers and volatile species. A grade selected only for oxidizing steam may be vulnerable during an incomplete-combustion excursion; a grade selected only for dry reduction may be poorly matched to the normal steam-rich operating state.

3. Heat-Release Geometry

Hydrogen and natural gas do not automatically create the same flame shape in the same burner. Burner replacement, nozzle geometry, velocity, swirl, dilution, oxygen enrichment and furnace recirculation affect where heat is released and how the flame approaches the lining or product. A shorter or differently anchored flame can move the maximum heat flux toward a burner tile, roof sector, sidewall or skid area.

Refractory temperature is not identical to gas temperature. Surface emissivity, product loading, view factor, deposit condition and convection determine the hot-face response. Qualification must therefore compare local refractory heat flux and temperature gradients, not merely set the same furnace-temperature target before and after conversion.

4. Gas Velocity and Recirculation

Changes in oxidant choice, dilution and flue-gas recirculation can alter gas volume and velocity. These changes affect convective transfer, dust entrainment, deposit location, erosion, joint pumping and the delivery of reactive species to the hot face. An atmosphere that is chemically mild in a static test may become damaging when high flow continually renews the reactant and removes the altered surface.

5. Product and Scale Chemistry

The product inside the furnace is part of the lining exposure. Research on hydrogen-fuel mixtures in reheating conditions shows that the changed atmosphere can influence oxide-scale formation on steel. That matters to the refractory because scale quantity, adhesion, melting behavior, fall-off location and impact can modify floor loading, hearth chemistry, skid deposits and downstream abrasion.

In other furnaces, the corresponding secondary effect may be glass volatilization, ceramic carryover, lime dust, reduced ore fines, metal splash or altered slag. A fuel-conversion study that tests clean refractory coupons without the process material may omit the dominant contact mechanism.

6. Startup, Trip and Shutdown Behavior

Steady full-load combustion is only one state. Purging, cold ignition, blend changes, low-fire holds, emergency trips and restart sequences may create the largest temperature gradients or the longest exposure to partially mixed gas. Condensation risk also rises as surfaces cross local dew points during cooling. The lining qualification must include transients because repeated small excursions can control campaign life even when full-load operation is stable.

Build an Exposure Atlas Before Discussing Grades

The most useful project document is an exposure atlas: a furnace drawing divided into zones, each with a defined thermal, chemical, mechanical and operational envelope. It prevents the phrase “the furnace atmosphere” from hiding local differences.

Burner Quarls, Tiles and Flame-Facing Roof Sections

These zones should be evaluated for maximum surface heat flux, thermal gradient, local reducing excursions, flame impingement, vibration, burner-tile geometry and joint movement. Burner conversion may change the temperature and stress field even when the furnace setpoint remains unchanged. Candidate materials need hot strength, thermal-shock tolerance and a chemistry compatible with both normal and off-design combustion.

Working Sidewalls and Load-Bearing Hearth Areas

Here the lining may experience product scale, rolling or sliding contact, mechanical load, deposit infiltration and long dwell time. A reheating furnace refractory decision must include the steel grade, scale regime, discharge temperature, residence time, hearth design and cleaning method. A fuel trial that changes scale formation can change the apparent refractory wear even if the refractory chemistry is unchanged.

Gas Exits, Recuperators and Waste-Heat Interfaces

These cooler zones may receive volatile condensates, dust and moisture. They can experience deposits that restrict flow, attack joints, upset heat recovery or create under-deposit corrosion of metallic components. The refractory may not reach the maximum furnace temperature, but chemical concentration and thermal cycling can make the zone life-limiting.

Doors, Penetrations, Expansion Joints and Anchor Interfaces

Air leakage and cold bridges make penetrations chemically and thermally distinct from the adjacent wall. A high steam concentration combined with a cold local surface can create condensation where a bulk gas calculation predicts none. Joints can also provide fast paths to backup insulation and anchors. The atlas should therefore map seals, shell attachments, door cycles, peep sights, burner penetrations and known leakage points.

Insulation and the Cold Face

It is unsafe to assume that only the hot face changes. Higher water-vapor loading, pressure fluctuations and altered permeability can affect vapor transport through joints or cracks. If moisture reaches cooler layers, thermal conductivity, anchor condition and shell-temperature patterns may change. Cold-face inspection ports and drainage or venting provisions should be reviewed as part of the conversion.

Translate Each Zone Into a Reaction Hypothesis

A good qualification program does not ask a laboratory to “test in hydrogen.” It states the mechanism that could make the lining unacceptable and designs evidence to challenge that hypothesis.

Hypothesis A: A Bond or Minor Phase Is Reduced and Lost

Laboratory work on bauxite-based castables in hydrogen-containing atmospheres has examined reduction and loss associated with iron-, phosphorus- and titanium-bearing constituents as well as changes to matrix phases. This illustrates why bulk oxide chemistry alone is insufficient. Two products with similar Al₂O₃ content can contain different reactive impurities, bonds and glassy phases.

The test should measure more than weight change. Phase analysis, microstructure, elemental mapping, dimensional stability, strength and fracture behavior can show whether a thin altered layer is harmless, whether the matrix is being depleted, or whether reaction paths connect to open porosity.

Hypothesis B: Steam Accelerates Oxidation of a Carbon-Containing System

Carbon-containing refractories may be selected for slag resistance and thermal properties, but their performance depends on oxidation protection and atmosphere. Steam is a potential oxidant at elevated temperature. The relevant experiment must reproduce carbon grade, antioxidants, temperature, gas flow, steam content, oxygen leakage and process contacts. An air-oxidation test alone may not represent the new service.

Hypothesis C: A Volatile Species Leaves the Hot Zone and Condenses Elsewhere

Reduction or steam reactions can generate gaseous species under some conditions. The resulting mass may not disappear from the furnace; it can recondense in cooler zones, react with deposits or change insulation behavior. Testing only the hot coupon misses the receiving surface. A gradient furnace or coupled hot-zone/cold-zone experiment can better reveal transport.

Hypothesis D: The Chemistry Is Acceptable but the New Heat Flux Cracks the System

A material can be chemically stable yet fail because the converted burner changes gradient, restraint or cycling. The test program should include thermal-shock or thermal-cycling work with a geometry representative of the lining, including joints, anchors or multilayer interfaces where relevant. Small unconstrained bars may overstate system tolerance.

Hypothesis E: The Refractory Survives but the Product-Refractory Interface Changes

More scale, a different scale composition, changed slag oxidation state or altered dust can create a new contact liquid or abrasion regime. In that case the correct test is not gas-only exposure. It is a coupled atmosphere–product–refractory experiment. The slag–refractory reaction-path framework is useful here because compatibility depends on wetting, penetration, dissolution, reaction products and removal—not on a family label.

The Pre-Switch Evidence Stack

Rather than approving one supplier certificate, the project team should assemble six connected records. Each record answers a different decision question.

Record 1: The Baseline Furnace Passport

Document current fuel composition, burner settings, excess oxygen, pressure, temperature distribution, product mix, throughput, scale or dust generation, shell thermography, lining design, repairs, campaign age, failure locations and startup procedures. A future comparison is credible only if the pre-conversion condition is known.

Record 2: The Converted Combustion Envelope

For every planned fuel step, calculate or model the expected dry and wet gas composition, water-vapor partial pressure, oxygen potential, adiabatic flame tendency, gas volume, burner momentum, heat-release pattern, wall heat flux and flue-gas recirculation. Include credible off-design states rather than one nominal point.

For a hydrogen oxyfuel furnace, oxygen purity, recycle-gas rate and dilution strategy are essential inputs. Oxyfuel should not be treated as air firing with nitrogen removed from a spreadsheet; the change can affect gas volume, radiative behavior, local temperature and equipment duty.

Record 3: A Materials and Construction Map

Record grade, batch, chemistry, density, porosity, bond, installation method, thickness, backup, anchor alloy, joint location, repair history and actual as-built deviations for every zone. This map allows field observations to be connected to a known material and exposure rather than to a generic furnace location.

Record 4: A Mechanism-Based Laboratory Dossier

The hydrogen refractory testing dossier should state sample preparation, pre-firing, atmosphere composition, dew point or steam addition, pressure, flow, temperature, heating and cooling rates, exposure duration, process-material contact and post-test analysis. It should also explain why each condition represents a furnace scenario.

At minimum, compare the incumbent lining with candidate materials under the same test. Reference samples in air or the natural-gas combustion atmosphere help distinguish normal thermal ageing from hydrogen-related changes.

Record 5: Instrumented Field Specimens

Laboratory screening should be followed by test bars, panels, tiles or coupons installed in mapped positions. Each specimen needs an ID, orientation, installation date, local exposure and planned retrieval interval. A specimen without a location and exposure history is not field evidence.

Field specimens are particularly useful when knowledge is incomplete. The TWINGHY and HYDREAMS initiatives illustrate the value of comparing refractory behavior as reheating furnaces move from natural gas toward hydrogen-containing or hydrogen-oxyfuel conditions. The industrial furnace itself becomes a controlled learning environment rather than a one-step commercial gamble.

Record 6: The Commissioning Decision Log

Record every fuel step, operating hold, alarm, trip, burner adjustment, product change and inspection result. Link the event to zone temperatures, gas composition, shell patterns and specimen condition. The log should preserve uncertainty and explain why the team continued, held, reversed or modified the trial.

Design the Laboratory Program Around Scenarios, Not One Maximum Temperature

A useful test matrix normally includes multiple atmosphere and temperature states. The exact values must come from the project, but the following logic is broadly applicable.

Representativeness Comes Before Severity

The most aggressive atmosphere is not automatically the most useful qualification condition. An unrealistically severe test may reject every candidate without explaining expected field behavior, while an easy nominal test may approve a material that cannot tolerate predictable transients. The matrix should include the intended operating state, credible deviations and a clearly identified accelerated condition when acceleration is scientifically justified.

Scenario 1: Natural-Gas Baseline

Expose the incumbent and candidates to the current combustion atmosphere and thermal cycle. This establishes normal ageing and prevents every post-test change from being attributed to hydrogen.

Scenario 2: Stable Hydrogen Blend With Complete Combustion

Reproduce the planned blend, steam fraction, residual oxygen and flow. This scenario evaluates the intended early commissioning stage.

Scenario 3: High-Hydrogen or Full-Hydrogen Steady State

Use the wet gas composition and thermal conditions expected at the highest approved conversion step. If the project includes air-fired and oxyfuel options, test them separately.

Scenario 4: Local Reducing Excursion

Represent credible incomplete mixing, low oxygen, high fuel, burner trip or purge-transition conditions. The purpose is not to invent an extreme accident but to challenge phases that may be stable only under the nominal oxidizing state.

Scenario 5: Cycling and Cold-Zone Moisture

Include repeated heating and cooling, low-temperature steam exposure or condensation where the exposure atlas identifies risk. Some degradation may accumulate at temperatures below the normal hot-face range.

Scenario 6: Process-Material Contact

Add representative scale, dust, slag, ore or product contact. Renew the contact material when industrial flow would continually refresh it. A static cup may screen reaction tendency, while a dynamic or rotating test may be required when renewal and removal control the mechanism.

Across these scenarios, useful outputs include mass and dimensional change, cold and hot strength, fracture energy, porosity and permeability, phase composition, reaction depth, elemental distribution, carbon loss, bond continuity and thermal properties. No single number is a universal pass criterion.

Keep an Unexposed and a Baseline-Atmosphere Reference

Every exposure series should retain material from the same production lot and include a reference aged under the incumbent furnace condition. Without those controls, normal firing shrinkage, specimen variability or pre-existing microcracks can be mistaken for a hydrogen-related mechanism.

Commission the Fuel in Evidence-Producing Steps

The fuel-switch procedure should be built around decisions rather than calendar dates. Each step needs entry criteria, a stable observation window, stop conditions and a defined next action.

Establish Comparable Operating Windows

Compare periods with similar throughput, product mix, furnace loading and discharge target. Otherwise, a change attributed to fuel may actually result from production. Normalize refractory and scale observations by operating exposure, not only by elapsed time.

Increase One Controlled Variable at a Time

Where practical, hold burner hardware, oxygen strategy, product family and throughput stable while changing the hydrogen fraction. If multiple changes are unavoidable, record them as a package and do not claim that the effect of hydrogen alone has been isolated.

Use Zone-Based Stop Conditions

A trial should pause for defined conditions such as an abnormal shell-temperature pattern, unexpected refractory surface temperature, unstable flame, excessive NOx, rising pressure drop, accelerated scale accumulation, deposit growth, joint opening, visible spalling or an unexplained shift in product oxidation. The U.S. Department of Energy notes that high-temperature hydrogen combustion can form NOx; burner and dilution strategies therefore remain part of the furnace conversion, not a separate environmental afterthought.

Inspect Before the Evidence Is Destroyed

Schedule visual records, thermal surveys and retrieval of field specimens at planned intervals. If a damaged lining is later removed, use a location-based sample architecture that includes hot face, reaction front, transition and unaffected reference. The site's refractory root-cause analysis guide explains why debris alone cannot reconstruct the initiating sequence.

What the Monitoring System Must Be Able to Prove

A dashboard is useful only when its measurements support a decision. For a conversion project, the minimum system should be able to answer five questions:

1. Which fuel and oxidant state was active? Preserve hydrogen fraction, natural-gas flow, oxygen or air flow, recycle gas, pressure, burner status and timing.
2. Where was the heat delivered? Use zone thermocouples, pyrometry, thermal imaging, product-temperature data or heat-flux proxies appropriate to the furnace.
3. What atmosphere reached the zone? Record residual oxygen, moisture or dew point where feasible, pressure and any relevant reducing-gas indicators.
4. How did the lining respond? Track shell temperature, hot-face observation, joint condition, deposit growth, dimensional scans, test-panel changes and repair consumption.
5. Did the process transfer a new burden? Track scale, dust, product quality, pressure drop, cleaning frequency, downstream deposits and maintenance labor.

Trend data should be tied to furnace coordinates. The digital refractory monitoring framework is relevant because it separates repeatable measurements and condition trends from an attractive but unvalidated prediction.

Rewrite the Purchase Specification Around Evidence and Limits

A useful hydrogen-ready refractory specification should not ask a supplier to check a box. It should define the service and request comparable evidence. The procurement package should include:

• furnace type, zone drawing and lining function;
• normal and transient temperature ranges;
• fuel blend steps, oxidant strategy and expected gas compositions;
• water-vapor, oxygen-potential, pressure and flow scenarios;
• product, scale, dust, slag or vapor contacts;
• cycling, mechanical load, abrasion and cleaning practice;
• lining geometry, thickness, joints, anchors and backup system;
• required laboratory methods and reference material;
• field-test design, inspection intervals and retained samples;
• pass, hold, review and rejection criteria;
• change-notification rules for formulation, raw materials or manufacturing route;
• technical responsibility for installation, dry-out and commissioning support.

The supplier should disclose which evidence is directly applicable, which is an analogy, and which remains a hypothesis. A test in 10% hydrogen at one temperature should not be presented as proof for full hydrogen-oxyfuel operation across an entire furnace.

Do Not Let Decarbonization Accounting Stop at the Burner

A hydrogen project may reduce direct fossil-carbon emissions while increasing refractory replacement, steel scale, cleaning, downtime, repair energy or product loss. Those downstream effects do not automatically negate the conversion, but they must be measured. The correct comparison combines fuel emissions with service performance.

For refractory fuel conversion, useful outcome metrics include campaign life, refractory mass consumed per tonne of product, repair frequency, rejected product, scale or dust yield, energy per tonne, startup losses and unplanned outage hours. These indicators connect lining reliability to the actual decarbonization result.

The performance-adjusted refractory carbon approach can support this comparison. A lower-carbon fuel combined with a shorter, unstable lining campaign may transfer emissions and cost rather than eliminate them. Conversely, a carefully qualified lining can enable the fuel conversion and protect its carbon benefit.

A Practical Release Decision

The project is ready to move to the next hydrogen step only when the evidence chain remains coherent:

• the actual fuel, oxidant and product state remained within the approved envelope;
• temperature and shell patterns show no unexplained zone shift;
• flame, pressure and emissions controls remained stable;
• refractory observations and field specimens show no accelerating damage mechanism;
• scale, dust, deposits and cleaning demand remain within agreed limits;
• all deviations have an owner and a technically defensible disposition;
• the next step has defined stop and reversal criteria.

If one of these statements cannot be supported, the correct action is not automatically to abandon hydrogen. It may be to change burner mixing, adjust oxygen or recycle gas, revise the ramp, improve sealing, relocate a sensor, modify a local lining zone, select a different bond system or extend the observation period. The purpose of qualification is to reveal which lever controls the risk.

Focused FAQ

Does hydrogen combustion always damage refractories more than natural gas?

No. The outcome depends on combustion mode, water-vapor concentration, oxygen potential, heat flux, gas flow, material chemistry, product contact and transients. Some incumbent linings may remain suitable; other zones may require redesign or a different material.

Is a dry hydrogen laboratory test sufficient?

Not for a combustion application. Dry reducing hydrogen can screen certain reaction pathways, but complete combustion may create a steam-rich atmosphere with residual oxygen. Both nominal and credible off-design states should be represented.

Which refractory oxide is best for hydrogen service?

There is no universal answer. Performance depends on phase assemblage, bond, impurities, porosity, temperature, atmosphere and contact material. The correct comparison is zone-specific and mechanism-based.

Why does the water-vapor partial pressure matter?

It helps define the chemical potential driving steam-related reactions and condensation. A percentage without total pressure and temperature is incomplete, and a hot-zone value does not describe a cooler joint or insulation layer.

Can the existing natural-gas burner simply run a hydrogen blend?

That is a burner-engineering and safety decision. Even where hardware permits blending, flame stability, velocity, heat release, NOx, controls and lining heat flux must be validated rather than assumed.

Should every furnace zone use the same new refractory?

Usually not. Burner tiles, roof, sidewalls, hearth, gas exit, doors and backup insulation experience different combinations of heat, atmosphere, product contact and movement. A zone map often supports targeted changes instead of a complete relining with one grade.

What is the most valuable field test?

An identified specimen or panel installed in a known zone, exposed through a recorded operating interval and retrieved for comparison with a baseline. Location and exposure history are as important as the laboratory analysis.

How long should a hydrogen trial run before approval?

There is no universal duration. The interval must be long enough to observe the proposed mechanism and include relevant production and cycling states. Early release may be possible for fast thermal responses, while chemical penetration or cumulative cycling may require multiple inspections.

What should be retained after the trial?

Fuel and oxidant records, burner settings, zone trends, thermal images, product and scale data, inspection photographs, removed specimens, reference materials, repair records, deviations and the signed release decision should be retained as one project dossier.

Conclusion: Qualify the Exposure, Then Qualify the Lining

Hydrogen conversion is not a material-substitution exercise. It is a furnace-system change that can alter atmosphere, steam loading, oxygen potential, heat transfer, product oxidation, deposits, cycling and cold-zone behavior at the same time. The lining must be judged against that combined service.

The strongest program begins with a mapped baseline, defines the new exposure by zone, states competing damage hypotheses, reproduces those hypotheses in laboratory scenarios, installs traceable field specimens and advances the fuel through measured steps. Procurement then buys evidence and operating limits rather than a generic “hydrogen-ready” claim.

That approach does more than protect refractory life. It helps the plant preserve safety, product yield, availability and the real carbon value of the conversion. The decisive question is not “Can this brick tolerate hydrogen?” It is “Can this lining system perform through every atmosphere, heat-flow pattern and transient that the converted furnace will actually create?”

#HydrogenFurnace #HydrogenCombustion #FurnaceRefractories #RefractoryQualification #HydrogenOxyfuel #IndustrialHeating #WaterVaporCorrosion #SteelReheating #FuelConversion #IndustrialDecarbonization