A purchasing manager in Manila sent me a bill of materials last year for a job that needed roughly 4,200 square metres of supply duct. The spec called for 25 mm double-wall panels throughout, "as per previous project". The building was a 24-hour call centre with a chilled water system running 6,000 hours a year and electricity at close to 0.20 US dollars a kilowatt hour. On those numbers, 25 mm was leaving money on the roof every single night.
I am not going to argue that thicker is always better. On a small shop office running one shift in a mild climate, 50 mm liner can take fifteen years to repay the extra panel cost, and nobody should sign that. The point is that the thickness decision is arithmetic, not habit, and the arithmetic is not difficult. Here is how I run it with buyers, using numbers you can check.
Every insulation has a thermal conductivity, lambda, in watts per metre per kelvin. Mineral wool runs about 0.035 to 0.040 at normal building temperatures, glass fibre similar, elastomeric foam around 0.036 to 0.040, and polyurethane or PIR board down at 0.022 to 0.026. Lower lambda means a thinner layer does the same job, which is why PU panels look so attractive until you price them.
Resistance is thickness divided by lambda. Add the surface films and you get the total R, and U is one over R. For a duct, the number that matters for the owner's electricity bill is U multiplied by the exposed perimeter and the temperature difference, which gives watts lost per metre of duct.
A 500 by 400 mm duct has 1.8 metres of perimeter per metre of length. A 1,000 by 500 duct has 3.0 metres. Same insulation, same U, but two thirds more heat loss per metre because there is simply more surface. This is why large ducts justify thicker insulation and small branch runs rarely do: the economics scale with perimeter, not with the length you happen to be counting.
Take a duct carrying air at 13 °C through a space at 25 °C, a 12 K difference, with mineral wool at lambda 0.038. Include an inner surface resistance of 0.13 and an outer of 0.04 for a duct in open air. Running the numbers:
At 3,000 running hours a year, the 25 mm duct loses roughly 78 kWh per metre per year. The 50 mm duct loses about 44. The difference, 34 kWh, is worth around 4 dollars a metre at 0.12 dollars per kWh. The extra liner costs perhaps 8 to 10 dollars per metre of duct once you account for 1.8 square metres of surface. Payback lands somewhere around two to two and a half years. That is a good number, and it is why I push 50 mm on almost any continuously running chilled system.
Now go from 50 to 75 mm. The saving drops to about 13 kWh a metre, roughly 1.6 dollars a year, against 9 dollars of extra material. Payback stretches past five years. From 75 to 100 mm it goes past ten. The curve flattens hard, and that flattening is the whole decision.
| Liner thickness | U-value (W/m²K) | Loss per metre at 12 K | Indicative panel cost | Typical payback vs previous step |
|---|---|---|---|---|
| 25 mm | 1.21 | 26 W | Baseline | — |
| 50 mm | 0.67 | 14.5 W | + 30 to 40% | 2 to 3 years at 3,000 h and 0.12 USD/kWh |
| 75 mm | 0.47 | 10 W | + 60 to 70% | 5 to 8 years, or 3 to 4 at 0.20 USD/kWh |
| 100 mm | 0.36 | 7.7 W | + 90 to 110% | 10 years and up, rarely justified on energy alone |
| 50 mm PU board | 0.42 | 9 W | + 70 to 90% | Depends on the board price; wins where space is tight |
Costs are material only, indicative, and will move with your local supplier. The paybacks assume the duct runs continuously; halve the hours and you roughly double the payback period.
Three situations flip the conclusion, and I have learned to check them before recommending anything.
Low running hours. A system that runs 1,200 hours a year has a third of the savings, so the same 25 to 50 step takes six or seven years instead of two. In a school or a seasonal hotel, 25 mm is often the right answer and I say so.
Cheap energy. At 0.04 dollars per kWh, insulation rarely repays on energy. Buyers in those markets should be choosing thickness for condensation control and acoustics, not for the bill.
Space and weight. A 100 mm double-wall panel on a 1,200 mm duct is a substantial object. Hangers, ceiling voids, and the structural allowance all change. Sometimes the honest answer is 50 mm plus careful sealing of the joints, because a well-sealed 50 mm duct outperforms a badly sealed 75 mm one by a wide margin. Leakage usually dwarfs conduction, and no amount of liner fixes a joint that was never sealed.
Panel and wrap are not the same product at different prices. A double-wall panel arrives as a rigid composite, cut and assembled like ordinary duct, with the insulation already bonded and the inner liner perforated or plain. Field wrapping means you fabricate bare galvanised duct on an auto duct line, then cut, fit and tape or pin mineral wool around it, then often add a vapour barrier and a jacket.
On straight runs of any length, panels win on labour every time — I would expect 8 to 12 square metres a day per pair of men on panel, against 4 to 6 on wrapped duct including the jacket. On awkward transitions, offsets and tight plantrooms, wrapping sometimes wins because you can work around obstructions that a rigid panel will not follow. If your fabricator has a duplex TDF machine, panels also get the same flange profile on both wall thicknesses, which keeps the joint detail consistent and saves a second setup.
One caveat on panels: the joint is where the insulation is interrupted. A 50 mm panel with a 50 mm thermal bridge at every flange is not delivering 50 mm performance overall. Corner inserts and flange gaskets matter as much as the liner thickness, and inspectors who understand this will look at the joints, and at the flange profile itself, before they look at the datasheet.
For chilled duct in a humid climate, the constraint is surface temperature. Moist air at 30 °C and 80 percent relative humidity has a dew point near 26 °C. Any surface below that will sweat, and sweating duct above a ceiling grid ends with stained tiles and an unhappy client.
In an open plantroom, the outside film resistance is around 0.04 and a 25 mm liner keeps the outer surface comfortably above dew point. In a still ceiling void the film resistance is closer to 0.11, and the surface runs colder. Recalculating with that value, 25 mm puts the outer surface at roughly 27 °C against a 26 °C dew point — technically above it and practically too close for comfort. At 50 mm you get around 28.5 °C, which is a margin I can defend to a client. That single change of assumption is why the same duct needs different insulation in two locations in the same building.
If the duct carries air below about 10 °C in a humid space, I would not go under 50 mm regardless of the payback maths, and I would insist on a continuous vapour barrier on the outside, with taped joints and no interruptions at hangers.
In the Gulf, the temperature difference across a supply duct in summer can reach 25 to 30 K, not 12. The losses scale almost linearly, so every step in thickness pays back roughly twice as fast as my base case. A 75 mm liner that looks marginal in Shanghai looks obvious in Riyadh.
In Southeast Asia the difference is smaller but the humidity is relentless, which pushes you toward thicker liner for condensation reasons and toward sealed, vapour-tight construction. In northern China the summer load is modest but winter heating ducts and fresh-air intakes running at minus 10 °C outside create the same condensation problem in reverse, on the warm side.
And in a temperate European market with 1,800 running hours and moderate energy prices, 25 to 40 mm is often genuinely correct. Buying 100 mm there is not virtue, it is just cost.
Give me four numbers: the duct perimeter or the largest size, the air temperature inside, the space temperature and humidity, and the annual running hours. Add your electricity price if you want a payback figure rather than a guess. With those four I can tell you where the knee of the curve sits for your building, usually within an hour.
Then check the joints. If the specification says 50 mm but the installation leaves flange bridges every 1.5 metres, you have paid for insulation you are not getting. Thickness is the visible decision. Sealing and continuity are the ones that show up on the energy bill.
Is 50 mm always better than 25 mm for chilled duct?
Not always, but it usually is on systems running more than about 2,000 hours a year with energy above 0.10 dollars per kWh. Below that, the payback stretches and 25 mm can be the honest choice. The exceptions are high-humidity spaces where condensation control overrides the economics.
How much does insulation actually save compared with duct leakage?
Often less. A duct system leaking 10 percent of its air is wasting more energy than the difference between 25 and 75 mm of liner. Seal first, then insulate. Buyers who upgrade insulation without fixing joints are optimising the smaller number.
Can I mix thicknesses in one system?
Yes, and you probably should. Thicker liner on long outdoor runs, main risers and cold ducts in humid voids; thinner on short branch runs inside conditioned space. Mixing is normal practice, as long as the specification says where each applies.
Does thicker insulation help with noise?
It helps, but not linearly. A 50 mm liner reduces breakout noise noticeably compared with bare duct; going to 100 mm adds less than you would hope. If noise is the driver, look at liner density and at the fan and fittings before you add thickness.