Cooling Water Contamination From a Tube Leak
Six shell-and-tube coolers in parallel cool a 92 wt% toluene stream from 95 C to 40 C on the tube side at 8.5 barg. Cooling water returns at 3.5 barg to a common header feeding one cooling tower. A VOC analyzer on that return header has climbed from 3 ppmv to 40 ppmv in four hours, and the cooling tower basin shows an iridescent sheen. Product quality is unaffected; the toluene analysis on the process outlet is unchanged. The plant wants to keep running for 36 more hours until a scheduled shutdown because the offsite customer has no inventory buffer. Each cooler has block valves on both sides, a vent and drain, and a sample point on the water outlet, but no online hydrocarbon detector. Cooling water flow to the bank is 480 m3/h. The process side runs 5 bar above the water side.
1. Determine the leak direction and explain the hazard each direction creates. Which is the immediate hazard here?
Read the worked answer
Direction
Process side 8.5 barg versus cooling water 3.5 barg. The process side is 5 bar higher, so the leak is process to water: toluene enters the cooling water, not the other way round. That matches the VOC analyzer and the sheen in the basin, and it explains why the product analysis is unchanged.
Hazards, ranked
- Hydrocarbons in the cooling tower. This is the immediate hazard. A 40 ppmv reading in the return header is a symptom of a liquid leak that will flash and strip in the tower. The tower has fans, motors and a basin now covered with a flammable sheen. The risk is a fire at the tower or in the basin, plus a VOC emissions exceedance.
- Loss of cooling water quality. The leak also carries process chemicals and can upset the tower chemistry and the biological treatment of the blowdown.
- Reverse flow on shutdown or upset. If the process side is depressurized while the water side stays at 3.5 barg, water will flow into the process. That contaminates the product and, more importantly, can create a water slug in a hot oil line. This is the hazard that is easy to miss because it does not exist while the plant is running normally.
- Leak growth. A tube or tube-to-tubesheet leak erodes; the rate will increase, not stay at 40 ppmv.
The point to make
The absence of a product-quality problem is not evidence that the leak is benign. The hazard has moved to a different part of the plant: the cooling tower and the basin.
2. Describe how you would localize the leak within two hours without shutting down the bank, and what you would do with the leaking exchanger.
Read the worked answer
Localize within the run
The bank has six exchangers with individual water outlet sample points. The fastest reliable method is:
- Sample each water outlet with a portable PID or a gas detector, or take a grab sample to the lab for toluene by GC. The leaking exchanger will read much higher than the other five. This takes about 30-45 minutes and does not disturb the process.
- Confirm by isolation if the sample points are not usable. Close the water inlet block valve on one exchanger at a time for 3-5 minutes and watch the return header VOC; a drop identifies the leaker. Before doing this, check that the process outlet temperature can tolerate the loss of cooling on that exchanger for the test period. With five others in service and the process outlet at 40 C, a short test is normally acceptable, but the downstream temperature must be watched.
- Do not use the process-side block valves for the test. Isolating the process side of a live exchanger creates a trapped, heating liquid volume and a thermal expansion hazard.
What to do with the leaking exchanger
Isolate it on both sides, depressurize the process side to the flare or a safe location, drain the water side, and confirm the leak with a low-pressure hydro test on the water side with the process side open and vented. Do not simply close the water and leave the process side pressurized: the leak will continue into the water side until the process side is depressurized.
If it is a tube-to-tubesheet leak and a spare bundle is available, plan the swap at the next opportunity. A plugged tube is acceptable only if the bundle has margin and the plug is installed properly. Do not put the exchanger back in service with the process side at 8.5 barg until the leak is confirmed repaired.
3. The operations manager argues that because product quality is unaffected, the plant can run 36 more hours. Give your response, including the specific limits you would set and the conditions under which you would escalate to a shutdown.
Read the worked answer
The argument
The operations manager is measuring the wrong thing. Product quality is unaffected because the leak is in the direction process to water; that is exactly why the hazard is outside the product specification. The relevant indicators are the return-header VOC, the basin sheen, and the LEL at the tower.
My response
I would not accept a blanket 36-hour run. I would accept a controlled run to the shutdown only under all of the following:
- The leaking exchanger is identified and isolated. The bank runs on the remaining exchangers, with the process outlet temperature and the water flow checked.
- The return-header VOC is trended and a limit is set, for example 100 ppmv, with a lower alarm at 50 ppmv.
- The basin is skimmed and the tower area is checked for LEL. If the LEL at the fan inlet exceeds 10%, the bank is shut down.
- No hot work or ignition sources in the cooling tower area, and the water treatment is adjusted for hydrocarbon contamination.
When I escalate
I shut down immediately if:
- The VOC continues to rise after the leaker is isolated, meaning there is more than one leak.
- The sheen reaches the basin in visible quantity or the LEL at the tower rises above 10%.
- The water flow to the remaining exchangers cannot hold the process outlet temperature, so the process side must be bypassed or the temperature rises above the downstream limit.
The 36-hour run is a calculated risk with monitoring, not a default. If the leak cannot be localized and isolated, the shutdown moves forward.