“Wet stacking” is a term that originally described a diesel engine dripping a thick, dark substance from its exhaust pipes or, as they’re often called, “stacks.” The dripping exhaust stacks were called “wet stacks,” and the engine was said to be “wet stacking.” The condition is caused by operating the engine at light load for extended periods, sending unburned fuel and soot into the exhaust system. Today, the term refers to an engine that isn’t completely burning all the fuel that’s delivered to its cylinders. Over a prolonged period, this condition can seriously degrade engine performance.
Most standby generators for facilities have a diesel engine as the prime mover. Many of these generators are routinely tested at no load or at light loads, for a variety of reasons. Building operators are reluctant to interrupt critical loads for transfer to generator and back to utility. Data center operators often refrain from switching uninterruptible power supplies (UPS) to emergency power during tests to avoid affecting battery warranties with excessive transfers. Generator sets may have been oversized in anticipation of load growth that didn’t materialize. Whatever the reason, diesel generators that aren’t regularly exercised at a significant fraction of their nameplate capacities are at risk for wet stacking.
Electrical design personnel, who typically make generator selections, are not always well-versed in the internal workings of diesel engines. The mechanisms responsible for wet stacking aren’t intuitively obvious. Consequently, wet stacking can be mysterious: It impacts generator selection, but it’s not well understood in the electrical profession.
The intent of this article is to demystify wet stacking by describing some of the fundamentals of diesel operation, how wet stacking happens in an engine and how it affects performance, solutions for existing facilities, and solutions for facilities in the design phase.
Diesel engine fundamentals
To understand how wet stacking happens, one must have at least a cursory knowledge of the operation of a diesel engine. Here’s a simplified description—looking at a single cylinder—of how a four-stroke, turbocharged diesel engine works:
The intake stroke: As the piston travels downward, the intake valve opens, and the turbocharger delivers compressed air to the cylinder.
The compression stroke: When the piston reaches the limit of its downward travel, it reverses direction, the intake valve closes, and the piston compresses the air in the cylinder. The temperature and pressure of the air in the cylinder rise dramatically.
Fuel injection: With the piston at the top of the compression stroke, the fuel injector sprays a fine mist of fuel into the cylinder. The air in the cylinder is hot enough to vaporize and ignite the fuel. The burning fuel adds heat to the air in the cylinder, and its pressure and temperature rise further.
The power stroke: The hot compressed gas in the cylinder pushes the piston downward. The force on the piston is transmitted to the crankshaft through the tie rod, turning the crankshaft.
The exhaust stroke: The exhaust valve opens, and the piston pushes the hot gas out of the cylinder, through the exhaust valve, into the exhaust system. A portion of the exhaust gas is diverted through the turbocharger to drive compression of the intake air, and the cycle begins again.
When a diesel engine runs without load, it develops only enough power to drive its accessories and overcome internal friction. It uses very little fuel and consequently develops little heat inside the cylinder. The cylinder’s internal surfaces stay considerably cooler than when running at higher load.
In the theoretical diesel cycle, the compression stroke is called “adiabatic.” That’s a thermodynamic way of saying that there’s no heat transfer between the air in the cylinder and its surroundings: the piston, cylinder head, and cylinder wall. In reality, though, the hot compressed air does lose heat to those surfaces. With no load, the cylinder’s interior is cooler than the engine’s design temperature, and the compressed air loses more heat to the engine than it would with the engine running under load.
A diesel engine doesn’t use spark plugs. It relies on the hot compressed air in the cylinder to vaporize and ignite the fuel. With the air cooler than the design temperature, conditions for combustion are less than ideal. The fuel ignites and burns, but it doesn’t burn completely. What remains are vaporized fuel and soot—small, hard particles of unburned carbon. Fuel condenses in the exhaust system, and soot deposits on surfaces inside both the exhaust system and the engine.
In the exhaust system, fuel vapors condense and mix with soot to form a dark, thick liquid that looks like engine oil. It may ooze from the turbocharger or drip from the exhaust outlets. The appearance of liquid on the exhaust stacks leads to the term “wet stacking.”
Inside the cylinder, soot can form hard carbon deposits on the fuel injector nozzle. The nozzle is designed to atomize the fuel, delivering a fine mist that vaporizes readily. When the injector nozzle is fouled with carbon, its ability to atomize the fuel is compromised, and it delivers larger droplets to the cylinder. The fuel consequently vaporizes less efficiently, more fuel remains unburned, and more fuel passes into the exhaust system. Wet stacking is a progressive condition; it tends to lead to more wet stacking.
It’s generally believed that prolonged operation at low loads can lead to permanent engine damage, requiring a major engine overhaul. Costs of an overhaul can run so high that replacing the unit is the most economical option.
NFPA 110-2010, Standard for Emergency and Standby Power Systems, paragraph 8.4.2, requires that units be tested monthly with adequate load to maintain the manufacturer’s minimum exhaust temperature, or at 30% or more of their nameplate rating, for at least 30 min. The explanatory material in Annex A states that these testing requirements are intended to reduce the likelihood of wet stacking. It’s worth noting that a provision in the 2005 version of NFPA 110 allowing testing at lighter loads was deleted in the current code. Building operators no longer have the option of testing generators at light loads.
The general cure for wet stacking is a few hours of operation at a load of about 75% of the generator’s nameplate rating or more, raising the exhaust temperature high enough to vaporize the unburned fuel in the exhaust system and blow out the soot. But, the exhaust temperature at that load is well above the auto-ignition temperature for diesel fuel, and on rare occasions fuel and soot can ignite within the exhaust system. If a unit has a history of extended operation at low load, or if there’s no documentation that it’s been exercised recently at adequate load, it’s important to have a professional generator maintenance expert manage the load testing procedure.
Existing facilities with inadequate available load for generator testing will have to find a method to comply with the requirements of NFPA 110-2010. For facilities with loads considered too sensitive to for the momentary interruptions required by generator testing, the low-cost alternative is to re-evaluate the sensitivity of those loads, and include them in the test protocol. An alternative with low first costs is to engage a professional maintenance firm to manage load testing, with temporary load banks as required to meet code requirements.
Possibly the least disruptive path to compliance is to measure exhaust temperature while powering the available building load. Many generators reach the manufacturer’s recommended exhaust temperature with loads less than 30% of their nameplate rating. This method requires measurement of exhaust temperature for each test, requiring the installation or purchase of measurement equipment, or the services of a generator technician. Compliance isn’t certain, but if the existing test load isn’t far from 30% of nameplate, this technique may well deliver compliance without significant changes to the facility, or to its existing test protocol.
A load bank with enough capacity to increase the total test load, including the facility’s minimum available load, high enough to reach either the recommended exhaust temperature or the prescribed 30% level, is adequate for compliance. This option requires a circuit breaker on the emergency system, a means for disconnecting the load bank if utility power fails during the test, and space for the load bank in a location where its heat can be safely rejected. The load bank may be permanently installed, a temporary bank owned by the facility, or a rental unit engaged for each test.
A facility may choose to connect nonessential loads to the emergency system to increase the test load, essentially using those loads in lieu of a load bank. This option will require one or more additional transfer switches or motorized circuit breakers, along with a load shed controller, to de-energize those loads if utility power fails. It may also require reconfiguration of the distribution to collect preferred loads on appropriate circuits. National Electrical Code 700.9(B)(5)c prohibits serving optional loads and emergency loads from the same vertical switchboard section. If a spare breaker position in an appropriate section doesn’t exist in the generator distribution, this option will require an additional section, switchboard, or disconnect.
All of the options available to existing facilities are applicable to new facilities. Generating systems should be specified with exhaust temperature monitoring, as this option is relatively inexpensive to provide with new units. Provisions for connection of a load bank adequate to test generators at 100% of their ratings, with automatic disconnection if utility power fails, should be part of any new installation. Nonessential loads can be connected to the emergency system, with appropriate load shed functions, at facilities where adequate generator loading is a concern. And, options exist at the design phase that are generally not economically feasible for existing facilities.
When a concern for adequate test load exists, the design team should consider using smaller generator units in parallel, with a “load demand” function. This function, implemented in the paralleling switchgear, compares online capacity to actual demand after the generating system stabilizes, and shuts off or restarts units as required to support the demand with a reasonable operating margin. Adequate test loads can be achieved with a fraction of the load required for a single-unit installation.
As an example, a system with two paralleled units can achieve 30% of a single unit’s capacity at only 15% of the total system’s capacity. The disadvantages of this approach include additional cost for paralleling gear and the space it requires, higher system complexity, and extended testing times required to test all of the units at adequate load. Code requirements for quick connection of essential loads will typically dictate the minimum size of paralleled generators at a level that’s adequate to support those loads with the smallest single unit.
A generating system that is capable of paralleling with the electric utility never lacks test load, as the utility acts as a nearly infinite sink for excess generator power. This option will generally be selected for reasons other than maintaining adequate test load, as it comes with a number of disadvantages. A facility may choose to parallel with the utility in order to do peak shaving, reducing its ratcheting demand charges, or to take advantage of payments from the utility for power delivered during system emergencies. Paralleling gear equipped for interconnection with the utility requires a number of functions that aren’t needed for island operation, including under- and overvoltage relaying, reverse power relaying, and redundant components specified by the utility. These functions significantly increase the cost of the paralleling gear.
Wet stacking is a serious condition affecting diesel generators that operate for extended periods at light load. It can be avoided by proper generator selection, and by properly performing routine generator testing.
Divine is project manager and electrical engineer at Smith Seckman Reid Inc.