Design Guidance

A shock needs a liquid slug that is driven fast into a closed end. The guidance on this page aims to prevent the slug from forming or moving, or to keep its speed low. It is derived from CFD studies of hot-gas defrost in ammonia refrigeration; the principles carry over to other systems.

When the risk arises

In ammonia refrigeration, hydraulic shock occurs most often in two places, both connected with hot-gas defrost:

• The hot-gas piping from the valve group to the evaporator, at the start of defrost. When the hot-gas valve opens, gas arrives at piping and coils that may still hold liquid, either as a layer along the bottom of the pipe or as a plug filling a low point.
• The wet suction piping from the evaporator, at the end of defrost. When the suction valve opens before the coil has been depressurised, liquid in the two-phase suction lines can be driven as slugs into sections of piping that are closed and have no exit.

The risk is highest during changes of operation: when a defrost sequence is interrupted, after a power loss, or when valves are opened by a timer without checking the pressure. The Mechanism page describes these situations in more detail.

Fast-acting and motorized valves

A fast-acting valve, such as a solenoid valve, opens practically at once. At that moment the pressure upstream of a hot-gas valve is around 100 psig, while downstream, in a low-temperature system, it is close to or below 0 psig. The flow through the valve is therefore choked: the valve passes the largest flow it can, and this initial surge lasts until the downstream pressure has risen.

A motorized valve opens over several seconds following its opening characteristic. The flow builds up gradually, and the downstream pressure has time to rise before the flow becomes large.

Two critical flows

Below a certain hot-gas flow no shock forms. Which critical flow applies depends on one question:

Can liquid collect in a trap upstream of the line?

No: liquid lies as a layer along the pipe. The gas has to raise waves on the liquid and lift it into a slug. The critical flow is given by the modified Taitel–Dukler criterion of Martin et al. (2007), explained on the Mechanism page:

For ammonia this is about 0.1 lb/s for a 4 in. pipe and about 1.0 lb/s for a 10 in. pipe.

Yes: a pre-formed slug may sit in a trap. The slug already fills the pipe, and the gas only has to push it. Its weight resists the push; if the gas flow is too small, the slug collapses under gravity before it can be driven along. The balance between the two is expressed by a densimetric Froude number,

where $U_g$ is the gas velocity, $\rho_g$ and $\rho_l$ the gas and liquid densities at the evaporating temperature, $D$ the pipe diameter and $g$ the acceleration due to gravity. With a critical value $Fr_{d,\text{crit}} = 0.01$, the critical mass flow is

Critical hot-gas mass flow (lb/min) for transporting a pre-formed slug in saturated ammonia. Calculated from Eq. (3) of IIAR Technical Paper 2025 RR1, with a critical densimetric Froude number of 0.01.

In the simulations, no shock formed at half the critical flow, and a shock formed in every case at twice the critical flow, for all pipe sizes and evaporating temperatures. At twice the critical flow the shocks stayed below 600 psia for pipes up to 10 in. The true critical value lies somewhere between these two limits.

The critical flow for a pre-formed slug is much lower than for a slug that has to form from a liquid layer: for a 4 in. pipe, about 2 lb/min against about 6 lb/min (0.1 lb/s). Where liquid can collect in a trap, the lower value governs.

Sizing hot-gas valves

Hot-gas valves are currently sized from the evaporator size and operating temperature, for a pressure drop of 2–5 psi across the valve (IIAR Refrigeration Piping Handbook, Table A8). This sizing considers steady flow only. It does not consider the initial surge when the valve opens and the flow is choked, which is exactly the moment when trapped liquid can be accelerated into a slug.

The studies lead to the following rules:

1. Flows of two or more times the critical flow produce shocks and should be avoided.
2. Size the first valve of a soft hot-gas arrangement close to the critical flow. It then passes enough gas to raise the downstream pressure, but not enough to create a shock.
3. The second valve may be larger, because the pressure drop across it will be lower once the downstream pressure has risen.
4. Do not energise the second valve until the pressure drop across it has been confirmed to be reduced, so that the flow through it does not exceed the critical flow.

Example: the Wiencke accident. The wet return header was 10 in., at –50 °F, and the critical flow for a pre-formed slug is about 19.5 lb/min.

According to the 2024 study, a ¾ in. first valve would have given no shock or only a light one, and would have raised the downstream pressure faster than the ½ in. valve that was installed. See Incidents for the accident itself.

Motorized valves

A motorized valve that opens slowly enough removes most of the shock. The simulations used the average opening characteristic of a common motorized valve type (ICM), with opening times from 0 to 16 s, for a 4 in. and a 6 in. pipe, each at two lengths.

Shock amplitude (psia) against valve opening time; h/D 50%, –25 °F, 1.93 lb/s. Source: IIAR 2022 Technical Paper #1, Table 9.

In a short pipe, where the slug reaches the end in well under a second, most of the shock is gone with an opening time of 1 s. A long pipe needs a longer opening time.

Rule for the opening time. The opening time should be well above the time the slug needs to travel the length of the pipework:

where $L$ is the pipe length and $Ju$ is the slug Joukowsky number from the shock correlation that fits the case: the 2022 correlation for a slug formed from a liquid layer, the 2024 correlation for a pre-formed slug in a trap. The rule was tested in simulations of both cases: four pipes with a liquid layer, and the Wiencke accident with a pre-formed slug. The study suggests a margin on the travel time of up to a factor of ten. For safety, the tools on this site use the full factor of ten, which gives the longer opening time.

Example: the Wiencke accident with a motorized valve. In the accident the slug was already formed in a trap, so the 2024 correlation applies. At –50 °F and 5.4 lb/s it gives Ju ≈ 3,860 psia; the simulation with an instantaneous valve opening gave 3,634 psia. Liquid ammonia at –50 °F has a density of 696 kg/m³ and a speed of sound of 1,812 m/s, so the slug velocity is about 20 m/s (65 ft/s). Over 17.5 m (57 ft) of pipework the slug needs about 0.9 s, which matches the time to shock in the simulation (0.92 s). With the factor of ten, the opening time should be about 9 s. Simulations of the accident with a motorized valve confirm the trend. They were run up to an opening time of 4 s, about four times the travel time, by which the shock had already fallen to 340 psia:

Shock amplitude in the Wiencke pipework against motorized valve opening time, at 5.4 lb/s. Source: IIAR 2022 Technical Paper #1, Table 11.

Evaporating temperature and pipe length

• Lower evaporating temperatures give stronger shocks. In the Wiencke pipework the shock fell from 6,000 psia at –60 °F to 380 psia at –10 °F.
• Longer pipes give stronger shocks (higher L/D). This matters for dead-end extensions of wet suction lines.
• A liquid layer around half the pipe diameter gives the strongest shocks. With a layer of a quarter of the diameter, the shocks were weak and erratic.
• Hotter hot gas gives stronger shocks, by about 13 psia per kelvin of hot-gas inlet temperature, because the gas enters faster at the same mass flow.

Checklist

• Avoid liquid traps in hot-gas and wet suction piping.
• Remove liquid from the coils as an intermediate step before admitting hot gas, interlocked with the controls.
• Keep the first hot-gas flow close to the critical flow; open a larger valve only after the pressure has been confirmed.
• Use pressure feedback (transducers or switches), not timers, both before opening the main hot-gas valve and before opening the main suction valve at the end of defrost.
• Consider motorized valves with an opening time well above the slug travel time calculated for an instantaneous valve opening.
• Make interrupted defrost sequences, for example after a power loss, return to a safe state.
• Take particular care at low evaporating temperatures and in long dead-end lines.

Other systems

The numbers on this page apply to saturated ammonia. The principles apply to any system in which a liquid and its own vapour share a pipe:

• keep liquid out of places that fast vapour can reach, or remove it first;
• limit the rate at which vapour is admitted, especially when a valve first opens;
• avoid closed ends where vapour can be trapped ahead of moving liquid.

In steam systems, the same conditions are set up mainly by the following:

• Warm-up and start-up. Live steam is admitted too quickly into cold pipework that has not been drained. The steam condenses on the cold walls, and the condensate collects in the line ahead of the steam.
• Failed or missing steam traps. Condensate is not drained continuously and builds up in the line.
• Pipe layout. Sagging lines, slopes in the wrong direction, and low points without drip legs (short collecting pockets that lead the condensate to a trap) let condensate gather where steam can reach it.
• Valve handling. Steam valves opened fully at once, instead of being opened a little at first so that the line warms up gradually; and flash steam from high-pressure condensate meeting cold condensate in return lines.

In depressurising liquid lines, such as a pulp-digester blow line, filling the line with liquid before the blow removed the source of vapour and with it the shock; re-routing the pipe alone did not.

The Tools page will provide calculators for the critical flows, the shock estimate and the motorized valve opening time.

References