Case study Climate tech · Data integrity
C-Layer
A bench-top rig that tests CO2-capture materials under cement-kiln flue conditions, tracks the health of every cartridge, and seals each result in a tamper-evident log.
- Self-initiated
- Sole designer
- Proposal and working 3D model, lab build next
The 3D preview needs WebGL, which is turned off in this browser. Everything else on this page works without it.
Figures come from the project proposal and its calculations. Real values are measured in the lab.
The question it answers
Which capture material survives cement-kiln smoke, and how long does it last before it loses its capacity? A plant has to know this before it invests in a full capture system. C-Layer answers it by measurement, not by guesswork.
The project does not claim new chemistry or better capture than commercial systems. Its value is measuring, accurately and honestly, what a plant needs to know.
The problem
- Cement is a big emitter. The industry released about 2.8 billion tonnes of CO2 in 2015, around 8% of global emissions. About half comes from the chemistry of turning limestone into clinker, not from fuel, so switching fuels cannot remove it. Only capture can.
- Kiln gas is a good target. It holds roughly 14-33% CO2, more than power-plant flue gas.
- Sorbents wear out. In one published study, PEI on silica lost 18%, 61% and 89% of its capacity after cumulative exposure to 10, 50 and 200 ppm of SO2, and 23% after 990 minutes in 10% oxygen.
- The right material depends on the gas. In the Svante and Lafarge pilot at a cement plant, the amine stage lost capacity while the MOF CALF-20 lost almost none.
- A local gap. We found no published data on Jordanian kiln flue gas after the dust filter. Collecting it is part of the plan.
The idea, simply
A capture material works like a sponge. It soaks up CO2 until it is full. Heat it and it lets the gas go, ready to soak again. Over time oxygen and sulfur in the smoke wear the sponge out.
To check its health, we pass gas through it and watch what comes out. At first no CO2 leaves, because the material takes it all. Near saturation it starts to leak through. That is the breakthrough curve. A fresh material holds out longer, a worn one fills faster, and the gap between the two curves is the measure of damage.
How the bench works
The rig is 80 × 33 cm and about 50 cm tall with its screen. Every part sits on a vertical board along a single gas line, left to right, with the P&ID diagram printed underneath in the same order. Anyone looking at it can follow the gas without an explanation.
CO2 and N2 cylinders → mass flow controllers → humidifier → mixing point → humidity sensor → valve V2 → valve V1 → cartridge or bypass → dryer → CO2 sensor → vent
Gas supply
Two mass flow controllers
Set an exact mix of 15% CO2 in N2 at 0.10 L/min. Nitrogen replaces air so no uncontrolled oxygen damages the material during a run.
Humidity
Split-flow humidifier
Part of the N2 bubbles through water and is mixed back to about 2% humidity, which a sensor measures instead of assuming.
Valves
V2 and V1
V2 switches between feed and a pure N2 purge. V1 sends gas through the cartridge or a bypass, so the same sensor reads the inlet and every reading can be divided by it.
Cartridge
The heart of the rig
A 12 mm stainless tube holding a 50 mm bed, clamped in a bolted aluminium block with two heaters. A thermocouple reads the middle of the bed, and an opaque jacket hides the material's colour.
Analysis
Dryer and NDIR sensor
A drying column protects an infrared CO2 sensor with a 0-20% range, which covers the 15% feed. Common low-cost sensors stop at 4%.
Control and safety
A separate control bay
A Raspberry Pi runs valves and heaters, reads the sensors and seals the log. A hardware thermal cut-off, an emergency stop and an extracted acrylic hood work independently of the software.
One test cycle
| Stage | What happens | Valves | Temperature | Time |
|---|---|---|---|---|
| 1. Pre-purge | Pure N2 drives air and moisture out of the material | V2 purge · V1 cartridge | 90-100°C | 10-15 min |
| 2. Cool | Cooled from outside, with no air on the hot bed | V2 purge | to 40°C | 5 min |
| 3. Inlet reading | Gas goes through the bypass to measure what goes in | V2 feed · V1 bypass | 40°C | 2 min |
| 4. Absorption | Gas flows through the bed and the sensor draws the curve to at least 90% | V2 feed · V1 cartridge | 40°C | 8-12 min |
| 5. Regeneration | Heat and N2 release the CO2 and reset the bed | V2 purge · V1 cartridge | 90-100°C | 10 min |
| 6. Compute and seal | Capacity ± error is calculated and the record is sealed | - | - | seconds |
The sealed data layer
This is where my cybersecurity major comes in. Every measurement goes into a chained, signed log, so any change after recording shows up at once.
| Part | How it works | What it guarantees |
|---|---|---|
| Record | Cartridge ID, time, raw readings, capacity ± error, heating energy, calibration and operator | Full traceability for every result |
| Hash chain | Each record carries the SHA-256 hash of the one before it | Changing any number breaks everything after it |
| Secure element | Records are signed with ECDSA P-256 by a key inside a chip such as the ATECC608, not in the Pi's memory | The key cannot be copied by software |
| External timestamp | A periodic RFC 3161 timestamp from a third party | The whole chain cannot be quietly rewritten |
| Calibrations in the chain | Reference-gas runs are stored in the same log | Every measurement links to a known calibration |
The honest claim: the log reveals any edit made after recording. It does not prove the sensor was right. That is the job of the calibrations and an independent TGA reference.
The blind test for the judges
- Four cartridges, opaque covers. A third person shuffles them, so the chance of ranking them right by luck is 1 in 24.
- Live curve. One goes in and its breakthrough curve draws on the screen over the saved curves.
- The call. The rig announces its capacity ± error and its rank. Then we lift the cover.
- Tamper check. We change one number in a copy of the log, and verification flags exactly that record.
Accuracy comes from an N2 balance for the outlet flow, a blank run to remove dead volume and sensor lag, at least three repeats per cartridge, and a TGA cross-check on some samples.
What is new
The innovation is not in the chemistry. It is in how the measuring, tracking and trust come together in one cheap, repeatable rig.
| Element | Status | Note |
|---|---|---|
| Breakthrough curve and capacity | Known method | Used as it is |
| PEI on silica, Zeolite 13X | Published materials | No new chemistry |
| Low-cost rig with a cement protocol | Contribution | 15% CO2, humidity, controlled O2 and SO2 ageing |
| Health history per cartridge | Method | Capacity and uptake rate across cycles, with error |
| Blind test with opaque covers | Method | Removes bias from the ranking |
| Signed, timestamped log | Contribution (security) | Trustworthy, auditable data |
Technical data sheet
| System type | Bench-top breakthrough rig for solid CO2 sorbents |
|---|---|
| Footprint | 80 × 33 cm, about 50 cm tall with the screen |
| Cartridge | Stainless tube, 12 mm inner diameter, about 50 mm bed, 2.5 g |
| Feed gas | 15% CO2 + 85% N2, 2% humidity, 0.10 L/min (0.669 mmol CO2/min) |
| Temperatures | Absorption at 40°C (PID), regeneration at 90-100°C in N2 |
| Heat-up time | 2-3 minutes, estimated |
| Sensor | NDIR CO2, 0-20%, for example GSS SprintIR-WF-20 |
| Controller | Raspberry Pi with a secure element, 7-inch screen |
| Materials tested | Zeolite 13X as reference, PEI on silica, CALF-20 if available |
Limits and next steps
Stating the limits plainly is part of the design, because every number here can be checked.
- Synthetic gas. No NOx or dust, and SO2 tests only with a supervisor's approval in a fume hood.
- Lab scale. Industrial heat transfer and plant design are out of scope. No carbon credits and no claim of net capture.
- Next: secure lab access with a supervisor, price the parts locally, finalise the model with real part sizes, write the sealed log (this can start now), run Zeolite 13X first to prove repeatability, then prepare PEI.
The full bench is modelled in 3D, with a live instrument panel, the six-stage cycle and a working signed log in the browser. Open the interactive model (in Arabic).