Counterintuitive but true: natural gas gains gauge pressure on the way up. How big the effect is, when it decides the checks, and why with LPG the sign reverses.
The pressure we measure in networks is gauge: the difference between the pressure in the pipe and the atmosphere just outside. Climbing, both fall — but not at the same rate. The outside air column «weighs» in proportion to the density of air; the gas column in the pipe weighs in proportion to the density of the gas. If the gas is lighter than air, as natural gas is, atmospheric pressure falls faster than the internal one: the gas's gauge pressure increases with elevation.
For typical grid natural gas (0.724 kg/Sm³ against air's 1.225) the coefficient is 4.91·10⁻⁵ bar/m ≈ 0.049 mbar per metre. Not a fine-engineering detail: it is the buoyancy of the gas column — the same physics as a hot-air balloon.
Case 3 of the validation dossier: a 300 m pipe in PE DE63, 20 Sm³/h, upstream at 25 mbar, customer 60 m higher.
The same elevation downhill would have taken 2.95 mbar: against an 18 mbar delivery requirement in low pressure, that is the difference between a passed check and a non-conformity.
The coefficient c changes sign with (ρair − ρgas). LPG is denser than air: climbing, the gauge pressure decreases; descending, it increases. In hilly LPG networks the effect penalises exactly the customers that methane-trained intuition would call advantaged — one more reason not to apply «from memory» a correction devised for a different gas.
The correct way to handle elevations is inside the flow law, not as an afterthought: the calculation works in the corrected variable ŷ = (P − c·(zelev − zmean))², so every pipe sees the buoyancy consistent with its end elevations, and elevations are referred to the network mean because only the differences have a physical effect. On a meshed network the effect couples with the flow split: a branch «helped» by elevation steals flow from the other, and only a simultaneous solution reconciles them.
Gasnetics applies the correction automatically to any mixture — sign included — and fills in ground elevations by itself as you draw the network on the map. The sizing accounts for it without the designer having to remember.
Because gauge pressure is measured against the local atmosphere, and climbing, air «weighs» more than the gas: the outside air column loses pressure faster than the gas column in the pipe. The net result, for natural gas, is a gain of about 0.05 mbar per metre of climb.
Whenever the elevation differences are comparable to the pressure margins. In low pressure a few tens of metres suffice: 60 m of elevation are worth almost 3 mbar, against a delivery requirement of 18 mbar. On hilly or mountainous networks the effect can decide the checks one way or the other.
No, it reverses: LPG is denser than air, so climbing the gauge pressure decreases and descending it increases. The coefficient changes sign with the sign of (ρair − ρgas), and a correct calculation handles it with the same formula for any mixture.