WSU Tree Fruit Research & Extension Center

Postharvest Information Network

Sunday, March 24, 2019

WSU-TFREC/Postharvest Information Network/Ammonia Leak Detection in CA Storage

Ammonia Leak Detection in CA Storage


The detection of ammonia refrigerant leaks before damage occurs to the stored product is a combined effort of three components:
  1. The ammonia detector and alarm system
  2. The sampling system
  3. The operator

As in any triangular relationship, if one component fails, the system fails. For example, a properly designed sampling system connected to an ammonia detector that has not been serviced correctly will not allow the operator to take corrective action. A poorly designed sampling system can render a functional ammonia detector useless. An operator who does not understand the limitations of detectors and sampling systems can misinterpret alarms.

Types of Ammonia Detectors

Ammonia detection is not ammonia analysis. Detectors are designed to indicate the presence of a substance at a level that can cause damage to people or products. They are not designed for absolute measurement for process control.

The most widely used ammonia detector in the apple CA industry is the Ultra-Violet (UV) detector (Honeywell, North Valley Industries). A UV light source is passed through the air sample where ammonia is possible. Since ammonia absorbs UV light a detector bulb senses the change in light.

The currently preferred routing of the sample is from the top of the sensor tube to the bottom. This flow is opposite of the original configuration and decreases the response time of the instrument.

The recommended minimum sample flow rate is 20 SCFH. Typical response time is approximately 6 minutes. In ammonia free air, CA and common storage will show a difference of 7 to 15 ppm, with CA indicating a lower reading. High concentrations of ethylene (or a poorly tuned forklift exhaust) will cause false positive readings. The recommended service interval is monthly with a possibility of as much as 6 months. The instrument should be warmed up for one week in advance of each season to eliminate false readings. After replacement of bulbs, an adjustment period of 48 hours is typically required. Sample inlet filters are required for rooms using lime for carbon dioxide control and are recommended for all other rooms.

Electrochemical is the next most common type of ammonia detector (Sensidyne, David Bishop Instruments). This system consists of an electrolyte fluid and a porous membrane. The membrane contains the electrolyte fluid while allowing the sample gas to pass through. When ammonia contacts the electrolyte, a voltage is created which is amplified and converted to a ppm reading.

The recommended sample flow rate is 1 LPM (2.1 SCFH). Typical response time is approximately 4 minutes. This response time can vary widely, based upon how the instrument is serviced and calibrated. There is no low oxygen effect noticed in this type of sensor. Ethylene and engine exhaust does not cause false positive readings. The recommended service interval is monthly. The instrument should be warmed up for 1 hour after a service is performed before calibration. This type of system can fail without indication of a problem, making maintenance very important. Sample inlet filters are not required but are highly recommended for all rooms being sampled.

Electrochemical detectors can also be disposable (Sensorstik). The electrolyte is sealed within the sensor and service consists of replacement of the sensor. The most common brand of disposable sensor used in the CA industry can operate down to 1.5% oxygen but only if it is calibrated in a low oxygen condition. The sensor should be stabilized for 48 hours in the low oxygen condition before calibration.

The Alarm System

The alarm system generally consists of a phone dialer, bell, and solenoid valves. The potential problems with this part of the system include: programming errors in the dialer, tone/pulse phone line incompatibility, and voice dialer/numeric pager incompatibility. Some systems connect the detector alarm circuits to the computer system, which is in turn connected to the alarm system. Other systems connect the detector alarm circuit direct to the alarm system. In either case it is recommended that a complete check of the system be made at least every year.

The Sampling System

The frequency of sampling for each room is critical to the proper detection of low-level ammonia leaks. The most vulnerable period for ammonia leaks is during a defrost cycle, due to the changes in pressure within the system. If the time period between samples of a room is longer than the length of a defrost cycle, then the risk of missing a low-level leak greatly increases. For example, if a room is sampled for 5 minutes every two hours and the defrost cycle is 20 minutes long, the chance of sampling during a defrost cycle is about 17%. If the time period between samples is decreased to 45 minutes then the chance of sampling during a defrost cycle becomes 44%.

If existing computer systems could be reprogrammed, then the ideal system would monitor each room as it enters a defrost cycle, for the duration of the cycle. When the defrost cycle is over, the sequence system could resume on the other rooms.

Detector response time determines how long each room needs to be sampled. If a detector can reach an alarm condition when exposed to a low level ammonia concentration within 3 minutes, then each room will need 3 minutes sampling time plus the time needed to pull a fresh sample. If three way solenoids are used, so a fresh sample is available when the sequencer selects each room, then no additional time is necessary.

The rate of sample gas flowing through the analyzer can also affect response time. It is important that the manufacturer's recommendations be followed regarding sample flow, as improper flow can make the detector nonfunctional or make more frequent service necessary.

The type of sample tubing may alter the concentration of ammonia that reaches the detector. Some absorption of ammonia is possible by tubing that would cause a lower than actual reading at the detector. It appears that this ammonia can later be released to the sample when the gas in the tubing is free from ammonia. The use of a continuous flow system where each tube has constant flow should minimize the effect of ammonia being absorbed into the tubing.

Ammonia detection systems using the UV detector require a much higher sample volume than other types of detectors. This large sample may need to be returned to the CA room in cases where the room gas-tightness is questionable.

The Operator

Proper operation and maintenance is critical for successful ammonia leak detection. Understanding the limitations to ammonia detection is also very important.

There are numerous factors which make low level leak detection difficult. Generally these conditions dilute the sample, which result in the detector operating properly, but indicate a lower concentration than that existing in the room. Some of these factors include the following:

  1. Nitrogen purge
  2. Humidifier systems
  3. Point of sample draw

If the CA room is under a nitrogen purge type system, this could easily dilute the sample, especially if the point of sampling is near the injection of nitrogen. Any water present in the room from defrost, humidifier systems, or flooded floors would also result in a sample at the detector lower than the concentration seen by the stored product.

These factors must be taken into account when interpreting alarms or minor indications of ammonia. A procedure should be established so that these factors are dealt with when alarms occur. Secondary low-cost ammonia detection systems can be utilized to double check for the presence of ammonia when the primary system alarms. The most cost effective of these is the handheld color change detectors. A sample from the room is drawn through a glass tube containing a chemical. The chemical changes color in the presence of ammonia. These tubes have a shelf life and are available for other gases as well. The ammonia tubes can also be used for verifying calibration gas.

Regular inspections of sealed CA rooms will also help minimize the potential for undetected leaks. The proper placement of view windows helps in seeing minor ammonia damage before major damage can occur.

All detectors currently in use require a long period of time to clear an ammonia alarm. If the sequencing stops when an alarm occurs, the operator can clear the instrument before sequencing resumes. If the detector continues to sample after an alarm occurs, the next room sampled may also alarm.

If the detector sensed a low-level ammonia leak but had not yet reached the alarm point, the next room in sequence may alarm. This can be a problem when the ppm reading is not recorded by the computer system. No record of an ammonia reading would exist for the first room - the one that had the leak. In this case, the room prior to the alarm should be checked for ammonia.

It is very important that the operator understand how their particular sequence system/detector system works. A back-up ammonia detector should be employed when any leak is detected to further isolate the area effected.

A properly maintained detector and sampling system may allow lower ammonia alarm set points, resulting in better leak detection.

The advisability of releases of ammonia into the mezzanine or cold storage areas should also be evaluated. If the low oxygen vent fans draw their air from this area, then the potential exists to contaminate rooms via these fans. This causes alarms to appear in rooms that are not leaking, further complicating the troubleshooting process.

Insurance Coverage

Currently there are only a few companies involved in insurance coverage for CA warehouses. Due to the recent losses, one of those companies has decided to no longer offer ammonia damage coverage.

It is important to remember that insurance policies typically have a 10% deductible. This means that an average ammonia damage will cost the warehouse approximately $20,000. Your insurance carrier is basing coverage on the fact that the detector is operational. If this is not the case and leaks go undetected, the deductible becomes a minor issue in the overall picture.

Additionally, this insurance is based upon your history of losses. Therefore a loss due to an undetected leak will impact your insurance rates into the future.

Independent calibration and maintenance is available, but it will be the warehouses and insurance carriers acting together to decide what is necessary for each situation. I strongly encourage you to determine, with your insurance company, what steps will be required.

Steve Houston, President

Pacific CA Systems Yakima, WA

14th Annual Postharvest Conference,
Yakima, Washington
March 10-11,  1998

Tree Fruit Research & Extension Center, 1100 N Western Ave, Washington State University, Wenatchee WA 98801, 509-663-8181, Contact Us