A waste heat recovery project is one of the very few capital decisions in a cement plant where the fuel is already paid for and the revenue is entirely predictable — and it is also one where published paybacks range from fourteen months to fifteen years for systems that look almost identical on paper. That spread is not measurement error. It is the gap between a design case built on annual averages and a plant that actually runs through a wet season, a six-stage preheater, and a kiln availability figure nobody put in the feasibility study. You can book a demo to see the twin run your own gas conditions before an EPC contract is signed.
WHR SIMULATION · DIGITAL TWIN · TURNKEY AI
Model the Boiler, the Turbine and the Export Before You Commit the Capital
iFactory builds a physics-grounded digital twin of your preheater and cooler gas streams, boiler duty, steam cycle, and turbine output — so retrofit sizing is tested against your real seasonal conditions and commissioning performance is verified against a model rather than a brochure.
35-40%
Of kiln thermal input lost as waste heat
25-45
kWh per tonne clinker recoverable
25-30%
Of plant electrical demand supplied
0.7-0.9 kg
CO2 avoided per kWh generated
The Available Heat
Grade Matters More Than Quantity
Roughly 35 to 40 percent of the thermal energy fed into a cement kiln leaves through the preheater exhaust and the clinker cooler air. On a plant producing a million tonnes a year that represents well over 60 megawatts of thermal energy going up the stack, and analyses of individual lines routinely put preheater exhaust at somewhere around 20 to 22 percent of heat input with the cooler contributing a further 14 to 15 percent. Those are large numbers, which is why the headline case for WHR is never in doubt.
What decides project economics is not how much heat is available but at what temperature. Thermal energy is only worth what the cycle attached to it can convert, and every hundred degrees lost in the gas path removes options permanently. The ladder below is the framework that governs cycle selection, boiler placement, and ultimately the megawatt figure that goes into the business case.
What Each Grade of Heat Can Actually Do
Highest grade
Cooler mid-tap gas
Mid-tapping the air quenching cooler draws gas at a meaningfully higher temperature than taking it downstream, which is why tap point selection is one of the highest-leverage design decisions in the whole system. This is the stream that justifies a steam Rankine cycle.
Supports: superheated steam generation, highest cycle efficiency
High grade
Preheater exhaust
The larger of the two streams by heat content, and highly sensitive to preheater stage count. Older four and five stage towers deliver hotter exhaust and therefore more recoverable power than a modern six stage tower, which is more fuel efficient but leaves less behind.
Supports: steam generation, typically two boilers per kiln line
Medium grade
Post-boiler and secondary streams
Gas leaving the WHR boilers still carries useful energy at temperatures where an Organic Rankine Cycle performs better than steam. ORC systems hold roughly 40 percent of the cement WHR market precisely because of their efficiency in this band.
Supports: Organic Rankine Cycle, low-temperature power generation
Low grade
Raw material and coal drying duty
The competing claim on waste heat, and the one that most often gets underestimated. Moisture in raw materials and fuel must be driven off, and that duty has first call on the gas — reducing what remains for power generation, seasonally and unpredictably.
Competes with: power generation for the same gas stream
Representative grading rather than fixed temperature bands, since actual values vary with kiln configuration, stage count, and tap point. The structural point holds everywhere: the drying duty sits between the boiler and the stack, and it is the variable that most often invalidates a static design case.
The last tier is where most feasibility studies go wrong, and it explains a striking amount of the payback spread in the published literature. Sizing a WHR plant is influenced directly by moisture content in raw material and coal, and moisture is seasonal. A system sized to average moisture conditions costs less to build but cannot recover the full available heat during the dry season, and during an extremely wet season the plant may have to run at reduced capacity. That is not a modelling nicety — it is the difference between a system that meets its guarantee and one that never does.
Why Projects Miss
Six Variables That Move the Answer More Than the Technology Does
Published payback figures for cement WHR range from as short as fourteen months at the optimistic end, through the commonly cited fourteen to thirty-six month band, out to studies reporting six to seven years and one Ethiopian case study arriving at fifteen years. The technology in those cases is broadly similar. What differs is the operating context, and the variables below are ranked by how much they move the outcome. A design case that fixes any of them at a single annual average is not a forecast — it is an assumption dressed as one.
Impact on Delivered Output Against the Design Case
Raw material and fuel moisture
Drying duty has first call on the gas. Wet season can cut recoverable heat sharply; dry season leaves capacity unused if the plant was sized to the average.
Preheater stage count
A six stage tower is a more efficient kiln and a worse WHR host. Four and five stage plants leave more recoverable heat, which inverts the usual efficiency logic.
Kiln availability and run factor
WHR generates around the clock only while the kiln runs. Every point of lost run factor removes generation hours the business case already counted as revenue.
Grid tariff and export terms
The single largest driver of the payback spread across published studies. Displacing expensive purchased power and exporting at a low feed-in tariff are entirely different businesses.
Cooler condition and tap point
Optimising grate cooler operation has been shown to raise heat recovery efficiency by over eleven percent. A degraded cooler quietly reduces the resource the boiler was sized against.
Boiler fouling rate
Dust-laden cement kiln gas fouls heat transfer surfaces continuously. Cleaning frequency assumptions made at design stage rarely survive contact with the actual dust loading.
Bar lengths indicate relative influence on delivered output against the design case, in both directions. They are a ranking device rather than a quantified sensitivity, since the magnitude for any specific plant is exactly what the twin is built to establish.
The tariff row deserves separate emphasis because it is frequently the only variable that is genuinely outside the plant's control. A WHR plant that displaces purchased grid power at industrial retail rates is competing against the highest price the plant pays for electricity. A WHR plant exporting surplus to the grid is competing against whatever wholesale or feed-in arrangement is available, which can be a fraction of that. The physical system is identical; the returns are not. Modelling the split between self-consumption and export across the load profile is therefore not a refinement — it is often the whole answer.
Model Architecture
What the Twin Actually Represents, Layer by Layer
A digital twin of a WHR system is only useful if it spans the full chain from kiln gas to electrical meter, because the constraints that decide output migrate between layers depending on operating condition. In the wet season the binding constraint is gas-side; at high ambient temperature it moves to the condenser; during a kiln upset it becomes steam quality and turbine trip protection. A model that stops at boiler duty answers the easy question and misses the ones that cost money.
Twin Layers From Kiln Gas to Grid Meter
Layer 1
Gas Side and Heat Source
Preheater exhaust and cooler air flow, temperature, and dust loading, driven by kiln feed rate, fuel mix, cooler condition, and stage configuration. This layer establishes what heat exists before any recovery equipment is considered.
Inputs: kiln feed, fuel, cooler performance, ambient
Layer 2
Competing Duty Allocation
The split between drying duty for raw materials and coal and the gas made available to the boilers, modelled seasonally rather than annually. This is the layer that most static feasibility models omit entirely.
Inputs: raw material moisture, coal moisture, seasonal profile
Layer 3
Boiler Duty and Fouling
Heat transfer performance across the preheater and cooler boilers, typically two preheater boilers and one cooler boiler per kiln line, with fouling progression modelled against dust loading so cleaning intervals become a scheduled variable rather than a fixed assumption.
Inputs: gas conditions, surface area, dust load, cleaning cycle
Layer 4
Steam Cycle and Turbine
Steam pressure, temperature, and mass flow through to turbine output, including part-load behaviour, which matters enormously because a WHR turbine spends a large share of its life away from design point as kiln conditions move.
Inputs: boiler output, condenser conditions, turbine curve
Layer 5
Electrical Balance and Export
Generated power against plant load profile, determining what is self-consumed at avoided retail cost and what is exported at whatever the applicable tariff allows, along with parasitic load from fans, pumps, and cooling.
Inputs: plant load profile, tariff structure, parasitic demand
Each layer feeds the next, and the twin is only predictive when all five are present. Layer 2 is the one most commonly missing from vendor models, which is a substantial reason why commissioned output so often falls short of the feasibility figure.
Layer four carries a subtlety that shows up in almost every operating WHR plant. Turbines are specified at a design point and quoted at design point efficiency, but a cement WHR turbine follows the kiln, and the kiln does not hold steady. Feed rate changes, fuel switches, cooler upsets, and planned rate reductions all move steam production away from the design condition, and part-load efficiency falls faster than most business cases assume. Modelling the actual distribution of operating points across a year, rather than the design point alone, typically produces an annual output figure several percentage points below the headline capacity.
TEST THE CASE BEFORE THE CAPITAL
Run Your WHR Design Against Your Own Seasonal Reality
Our team will build the twin from your kiln gas conditions, moisture profile, preheater configuration, and load pattern — then show you the output distribution across a full year rather than a single design point.
Cycle Selection
Steam Rankine, Organic Rankine, or Kalina
Three thermodynamic cycles compete for cement WHR duty, and the selection is genuinely temperature-driven rather than a matter of vendor preference. Steam Rankine dominates where gas is hot enough to justify it and is widely deployed in India and China at scale. Organic Rankine holds around 40 percent of the market on the strength of its low-to-medium temperature performance. Kalina remains the least deployed of the three. Book a demo to see the comparison run against your specific gas conditions.
The row worth arguing over is part-load behaviour, because it interacts directly with the kiln reliability picture. A plant with a stable, high run factor and a hot gas stream is a natural steam Rankine host. A plant with a six stage preheater, a variable kiln, and no existing steam operating competence has a much stronger organic cycle case than the raw efficiency comparison at design point would suggest. This is precisely the kind of trade-off a twin resolves quantitatively rather than through vendor argument, because it depends on the distribution of operating points at your plant rather than on the peak figure in a specification sheet.
Pre-Contract Decisions
Five Questions to Answer Before Signing the EPC Contract
The strongest argument for building a twin before a WHR project is that every question below has to be answered anyway, and the only real choice is whether it gets answered by modelling in advance or discovered during commissioning. Each carries a specific cost of being wrong, and in every case the error compounds because the capital is already committed by the time the answer arrives.
01
What megawatt rating actually matches our gas, in every season?
Sizing to average moisture conditions produces a cheaper plant that cannot capture full dry-season heat and may need reduced capacity in an extreme wet season. Sizing to peak conditions produces a plant that spends much of the year at part load. The twin resolves this by producing the annual distribution rather than a single number, so the capacity decision is made against the shape of the year.
Cost of getting it wrong: permanent oversizing or a structural shortfall against guarantee
02
How much heat can we take before drying duty is compromised?
Power generation and raw material drying draw on the same gas, and taking too much for the boilers means buying supplementary heat for the mill or accepting moisture in the feed. The boundary is seasonal and depends on quarry moisture, coal moisture, and mill throughput, all of which the model can vary together rather than assuming a fixed annual split.
Cost of getting it wrong: mill throughput loss that offsets the power gain
03
Where should the cooler tap point sit?
Mid-tapping the air quenching cooler gives access to hotter gas and therefore a better cycle, but it interacts with cooler heat recovery back to the kiln as secondary and tertiary air. Take too much, too hot, and secondary air temperature falls, which raises specific heat consumption on the kiln. The twin models both sides of that trade rather than optimising the WHR system in isolation.
Cost of getting it wrong: kiln fuel penalty quietly cancelling the power benefit
04
What proportion is self-consumed versus exported?
Displacing purchased power at industrial retail rates and exporting at a feed-in tariff are financially different outcomes from the same megawatt-hour. Modelling generation against the plant's actual load profile establishes the split, and that split frequently moves the payback calculation more than any equipment selection decision in the project.
Cost of getting it wrong: a business case built on the wrong price per unit
05
What happens to output as the boilers foul?
Cement kiln gas is heavily dust-laden and fouling of heat transfer surfaces is continuous rather than occasional. Design cases are typically quoted clean. Modelling fouling progression against real dust loading converts cleaning frequency into a planned operating variable and gives a realistic average annual output rather than a first-week figure.
Cost of getting it wrong: performance decline read as equipment failure
Question three is the one that most often surprises project teams, because it is the point where the WHR system and the pyroprocessing line stop being separate projects. Every unit of heat taken from the cooler at a high tap point is heat that did not become secondary or tertiary air, and secondary air temperature drives kiln fuel consumption directly. A WHR system optimised purely for generation can therefore raise specific heat consumption enough to erode a meaningful share of its own benefit — an outcome that is entirely predictable in a model and entirely invisible in a standalone WHR feasibility study.
Commissioning and Beyond
The Twin Earns Its Keep Twice
Most digital twin discussions stop at the design stage, which undersells the asset considerably. A calibrated model built for retrofit planning becomes the reference against which commissioning performance is judged, and then becomes the live benchmark that tells operations whether today's output shortfall is a fouled boiler, a wet feedstock, or a kiln running below rate. The comparison below shows how that reference is used across the asset life.
Three Stages, One Model
Stage 1
Retrofit Planning
Capacity sizing, cycle selection, tap point placement, and the self-consumption split tested across the full seasonal envelope before any specification is fixed.
Output: a defensible capacity decision with the year's shape behind it
Stage 2
Commissioning Verification
Measured output compared against modelled output under the same measured gas conditions, so a performance guarantee is assessed against the conditions that actually occurred rather than the contractual design point.
Output: an evidence base for guarantee discussion with the EPC contractor
Stage 3
Live Operation
Continuous comparison of actual against expected output for the current gas, moisture, and kiln condition, so a shortfall is attributed to a cause instead of being absorbed as normal variation.
Output: fouling alerts, degradation trending, and attributed loss
Stage two is where the commercial value concentrates on a retrofit project. Performance guarantees are written against a design condition, and the argument that follows a shortfall almost always turns on whether the plant supplied the gas conditions the contract assumed. Without an independent model, that discussion is a matter of competing spreadsheets. With one, both parties are looking at modelled output under measured conditions, which turns a dispute into a calculation.
The returns available justify the rigour. WHR systems can supply 25 to 30 percent of a cement plant's total electrical demand, have been associated with reductions in indirect emissions of up to 63 percent, and are reported to cut clinker production costs by roughly 3.8 to 7.5 percent. Generation potential sits in the range of 25 to 45 kilowatt-hours per tonne of clinker, and every kilowatt-hour generated avoids somewhere around 0.7 to 0.9 kilograms of carbon dioxide — which puts a six to nine megawatt system in the region of 30,000 to 55,000 tonnes of avoided CO2 per year. More than 70 percent of European cement plants have already adopted some form of waste heat recovery, and one analysis identified over 576 megawatts of ORC potential across 27 European plants alone.
Frequently Asked Questions
Cement WHR Digital Twin — Common Questions
Can a twin be built before we have a WHR system installed?
That is the highest-value moment to build one. The gas-side layers depend entirely on your existing kiln, preheater, and cooler, all of which are already running and already instrumented, so the heat resource can be characterised accurately from historian data before any equipment is specified. The recovery-side layers are then modelled from equipment characteristics rather than measurements, which is exactly what a feasibility study does — the difference being that the gas resource underneath it is measured across a full seasonal cycle instead of assumed from an annual average. You can
book a demo to see what your existing data supports.
How long does it take to build a model that is trustworthy enough to size capital against?
The gas-side characterisation needs enough operating history to cover the seasonal range, which in practice means a full annual cycle of historian data if it exists, or a shorter window supplemented with quarry and coal moisture records. Where a year of history is available the model can typically be built and validated within weeks, because the work is calibration rather than data collection. Where seasonal variation is extreme and history is thin, the honest answer is that the wet season figure carries more uncertainty than the dry season one, and the model should present that uncertainty rather than hide it inside a point estimate.
Our preheater is six stage. Does WHR still make sense?
It can, but the economics are different and the cycle selection usually changes. More preheater stages make the kiln more thermally efficient, which is desirable in itself, but it means less heat leaves in the exhaust and therefore less is available for power generation. Four and five stage towers have higher WHR potential precisely because they are less efficient kilns. On a six stage line the recoverable stream tends to sit in the temperature band where an Organic Rankine Cycle performs better than steam, and the sizing conversation shifts toward a smaller system with a stronger self-consumption case rather than an export case.
Does the twin interact with our existing control system?
It reads from it and does not write to it. The model consumes historian and DCS data to know the current gas conditions, kiln state, moisture, and load, and produces an expected output figure against which actual generation is compared. Control of the boilers, turbine, and electrical system remains entirely with your existing systems and operating procedures. The output is a reference and an attribution — this is what the system should be producing under these conditions, and here is which layer explains the gap — rather than a control action.
How does this help with a performance guarantee dispute?
By separating two questions that otherwise get argued as one. A guarantee shortfall can be caused by equipment underperforming or by the plant supplying gas conditions below the contractual assumption, and without an independent model both parties simply assert their preferred explanation. A calibrated twin computes expected output under the gas conditions that were actually measured during the test period, which isolates equipment performance from resource availability. Our team can walk through how that verification is structured for a specific contract through
support.
IFACTORY · CEMENT · DIGITAL TWIN AND TURNKEY AI
Know the Answer Before the Capital Is Committed, and Prove It After
iFactory models your WHR chain from kiln gas through boiler duty, steam cycle, and turbine output to the grid meter — sizing retrofits against your real seasonal envelope, verifying commissioning performance under measured conditions, and attributing every megawatt-hour of shortfall to a cause.
5 layers
Gas side through to grid export
Seasonal
Modelled across the full year, not a design point
Read only
Reads your DCS, never writes to it
3 stages
Planning, commissioning, live operation
