Fires and Explosions in the European dairy industry were becoming an increasing problem. In the early 1990s groups of people from industry, insurance providers, and suppliers of protection equipment came together to analyze events and case studies, to find a new approach to solve the persistent problem. Three hundred and fifty combined fire events were reviewed. Forty resulted in explosions with a majority attributed to product decomposition as the ignition cause.

While the data was the same the group in Germany decided that if they could prevent the event, they would not have to deal with it. A group in Ireland/UK decided it would not be possible to cover all causes, so they decided to develop a protection system that addressed the needs of the dairy industry.

Prevention of Ignition Source
Basically, all four necessary preconditions for an explosion can be present: fuel (milk powder) powder deposits and swirling dust, oxygen, ignition sources, decomposed milk products, confined space.

(see Photo 1)

Operating experience shows that the primary source of ignition was smoldering spots or self-igniting milk products. Even with the best safety measures and process controls in place, it is not possible to effectively prevent the deposit of milk products in spraying drying devices or air diffusers, nor can caking on the vessel walls be avoided. The danger exists that, through long-term exposure to hot air, a thermic decomposition of the deposits will be initiated and that this will lead to smoldering spots and/or self-ignition of the products.

Depending on moisture, fat content and on the air flow, smoldering nests in milk products will form solid and compact structures. Because of the bad diffusion of oxygen through the pores, the smoldering nests will expand from inside outwards rather slowly. Various conducted tests have indicated that small smoldering spots of milk products have quite low surface temperatures and therefore are not effective sources of ignition for dust-air mixtures. The low surface temperature makes them hard to see with standard IR sensor technology until they break apart, exposing the hot surface providing an ignition source.

As a rule, such compact glowing deposits will only become a source of ignition when they detach themselves and fall into the lower tower areas or are transported into secondary components where potentially explosive dust-air mixtures is present. Therefore, it is essential to detect smoldering material in an early stage, in order to be able to take appropriate measures. If a method could be developed to detect the growth of these deposits at an early stage the product flow could be stopped eliminating the hazard. The potential ignition source can then be dealt with manual and/or automatic means before product supply was returned.

An early recognition of smoldering fires at an initial stage is possible through inspection of the exhaust air from drying installations for the presence of carbon monoxide, a gas that is the product of the thermic decomposition of milk products. Because of the high air flowrate within milk powder drying installations, the produced carbon monoxide is diluted so strongly that an extremely sensitive measuring system is required.

(see Photo 3)

With the usual exhaust air volumes up to 100,000 cu m/hr, an increase of the CO content in the exhaust air of less than 1 ppm can be an indication of a smoldering spot but environmental contamination can be much higher. This problem could be solved by means of differential measurements between the air intake and the exhaust, where only the CO content produced in the drying apparatus is taken into consideration.

The characteristic of the heteroatomic gas CO, to absorb infrared light in specific bands between the frequencies 2.5 and 12 pm, is used in infrared spectroscopy to provide a means for determining concentration levels. With the Non-Dispersive Infrared Absorber (NDIR), a measuring principle is available that is suitable for detection of traces of carbon monoxide levels. NDIR CO gas analyzers, with a measuring range of 0 to 10 ppm.

To accomplish the measurement small gas samples would be continuously extracted from the drying apparatus and pumped through a measuring cell, which has been fitted with windows that permit infrared rays to penetrate. A ray of light, which is directed through the windows and penetrates the gas, is weakened in the area of certain frequencies, before it meets the detector. This absorption correlates with the CO concentration and is defined by Lambert-Beersch's law:

A = I₀-I = 1 - e ^{-e.c. l}
       I₀

A = absorption
I_o = incident radiation capacity
I = emission radiation capacity
e = extinction coefficient
c = concentration
l = length of cell

To perform this analysis in the time necessary both samples needed to be tested simultaneously. By using a so-called cross flow-modulation procedure the air intake of the drying apparatus is used as a reference gas. The exhaust air sample, which is to be measured, and the reference gas, is alternately introduced into the measuring cell through a micro solenoid valve.

Infrared light generated from the infrared source passes through the measurement cell and enters a detector containing the gas to be measured. When zero gas is sent to the measurement cell, more infrared light reaches the detector. On the other hand, when sample gas is sent to the measurement cell less infrared light reaches the detector. The degree of this infrared-light attenuation is related to the concentration of CO gas in the measurement cell.

An early-warning fire detection system, which is based on CO detection, consists of the following main components: gas sampling probe, sample gas preparation, analyzer, process system controller interface. To be acceptable as a prevention alternative a co early-warning fire detection system must provide quick recognition of smoldering fires while avoiding false alarm activations.

First the process system must be analyzed to determine the proper points of sampling to get a true picture of the systems process flows. In summary, probe locations must assure all air in = all air out is measured with respect to the CO content of the air.

Because the exhaust gases are loaded with dust and have a high dew point temperature, a dust and water removal system is required. The input air to the analyzer must first go through various air preparations to remove moisture from the line and balance the flows to the analyzer.

(see Photo 2)

To avoid a false alarm, caused by a sudden rise in CO content in the intake air (for instance as a result of heavy traffic), the transit time in the sampling lines must be balanced. The gas sampling lines to the analyzer should be kept as short as possible, in order to avoid unnecessary delay times. Note: This initially proved to be an insurmountable task when considering direct fired dryers.

A CO detection system was tested on an industrial level, during a test program organized by the "BG Nahrungsmittel und Gaststatten" in Germany. The system, consisting physically of an infrared gas analyzer, proved itself capable of detecting smoldering spots, which, because of their size, could not be effective as a source of ignition.

A comparison between the progression of CO content in the exhaust gas and the concentration noted from an artificially introduced smoldering spot was made. An increase from 0 to 1 ppm CO developed within seven minutes, within a further 16 minutes the CO content rose from 1 to 5 ppm, and thereafter the value rose within 15 seconds to 8 ppm. Four minutes later water was introduced, and after three hours of shut-down the installation could be used again. The test installation was not equipped with a dedicated alarm system at that time. If we assume that clear signs for a smoldering fire were already available below 1 ppm CO, then the operator lost 15 minutes of time, during which he could have initiated safety measures.

From the experience gained with the test installation it can be concluded that an unusually steep increase in CO content in the exhaust air can best be used to trigger a pre-alarm allowing the operator enough time to localize and remove the smoldering spot. The initiation of a forced shutdown, or the activation of fire extinguishing installations, would occur automatically, after a certain individually determined threshold has been exceeded.

From the testing a system was developed for the spray dyer application. The CO detection system consisted of an IP 54 housing, in which the sample gas pumps, the cleaning, conditioning units, gas analyzer and processor evaluation unit were installed and protected from the harsh operating conditions of the installation.

(see photo 8)

The unit calculates the differential value of the CO concentration, between the exhaust and the intake air of the drier. In this way, only the CO value created within the spray drying unit is used. The detection system makes use of a programmable logic controller (PLC) that permits a permanent supervision, as well as a gradient-oriented definition of the alarm threshold values.  

(see Photo 5)

The system can recognize the operating mode of the installation by means of an interface and can automatically activate a different set of threshold values, to prevent false alarms. If success was based solely on the ability of the system to prevent explosions, then the first-generation system exceeded expectation. But, unfortunately success lead into many new challenges. 

The first system modifications performed a volumetric correlation of vessel flows. Through a combination of calculation algorithms, flow control, flow mixing and volumetric adjusting components the behavior of the dryer could be compensated for and fine-tuned as required to meet field conditions. The new buffer system to adjust the real-time system flows to effectually allow the inlet sample and outlet sample to be analyzed at the same relative point in time. This reduces the potential of a stray CO signal entering the inlet and causing the indication of a CO rise because of the spray dryer volumes delaying effect on the flows.

With new increased filtering power, they have a system that can deal efficiently with the effects of most direct fired dryers. While they do create a challenge considering the CO fluctuations involved, most dryers can be provided with protection if the air flow from the burner is not laminar in nature but turbulent.

Typical Resultant Synchronization and Measurement Readings

(see Photo 4)

Protection
As indicated, a group of Irish and British researchers at the same relative point in time were looking at ways to deal with a developing deflagration. They determined the following needs for a protection system:

      1 – Keep deflagration overpressure to a safe non-destructive level
      2 – It must be good in large volume areas
      3 – No contamination form agent used
      4 – Actuator could not introduce contaminants
      5 – Easily maintained
      6 – Low-cost maintenance

(see Photo 7)

Since explosion suppression systems were used to protect industry successfully for years the next step was to review existing protection systems to determine if any were suitable or could be modified to be suitable for use. But existing dry chemical systems could not meet the large volume requirement without high pressures and used explosive actuators or gas generators that were a contaminant and, in most cases, required expensive factory maintenance. Finally, water was reviewed and, while meeting most of the requirements, did not have a delivery system available, plus the waters cooling effect on hot dryers could be as problematic as the deflagration itself.

Through research they understood that if water could be released in a droplet size below 50 micron an explosion could be suppressed. But limitations in pressure, energy cost, and flow presented major roadblocks to success and the small orifices required to reduce the droplet size were problematic and unreliable.

They then turned their attention to pressurized hot water, but it also had development concerns. First, it had to be shown to be reliable and efficient for the application. The operational parameters had to be defined and had to be tested for the required range of Kst and Pmax values.

To evaluate hot water suppression as a potential solution, a test program was developed and performed. Tests were performed in two different sized vessels with different geometrical characteristics. The first vessel was a 2.8 cu m ISO test rig. The second was a 28 cu m test vessel with a 7:1 aspect ratio to mimic a typical tower dryer. Future testing in a 140-cu-m verified the results.

The results indicated that droplets under 50 microns in size performed the best deflagration suppression. A comparison to other agents found water droplets under 20 microns to function as a total flooding system with extinguishing values twice that of the Halons and equal to or better than the dry chemical agents used for explosion protection.

With the tests a success a suppressant was found, water droplets under 50 microns. A review of pressure-heated water found that water heated to its boiling point increases its liquid heat content, temperature, and pressure. This surplus energy increases the amount of flash steam produced. In addition, when stored and discharged from a storage cylinder the pressurized hot water has a more constant discharge flow rate then a nitrogen pressurized cylinder.

Researchers noticed a major problem with dry chemical systems because they released like a shotgun at an ever-slowing rate of discharge. Pressurized hot water had an advantage, when released into the protected vessel it flashed to a vapor accelerating in a 360° pattern as it does so provide the ideal agent for a large volume application like a spray dryer.

Using available explosion detection and control equipment the only thing required was a hot water storage and deployment system. Development produced a dual lined stainless steel vessel with a water heater up the middle to produce a 180°C temperature resulting in a 10-bar pressure level.

The pressurized hot water system provided the following industry required protection needs:

1 – Reduced pressures (Pred) were acceptable for vessel protection.
2 – It was effective for all vessel volume ranges.
3 – Water is a food safe substance.
4 – A gas generator was used out of the product stream eliminating contamination.
5 – The system was designed to be easily maintained.
6 – By using water the maintenance was expected to be lower in cost.
7 – Equipment must be provided in stainless steel execution as with the spray drying facility they protect; standard mild steel suppressors would not be acceptable.     

Suppressors
The suppressors need to function as a storage system keeping the suppressant at the right temperature and pressure ready for activation. The cylinder was designed to eliminate the stress effects of the heat on the life of system components. The heater controls were remote connected to lower problem potentials and make user servicing simpler. The pressure regulator for temperature control was also remote mounted via a high-pressure flexible conduit to reduce heat effects. A new release valve was developed. The release system provided a reusable and removable valve that lowered the labor cost of system reconditioning and therefore the total operational cost.

(see Photo 9)

Deployment System
By using a 4-in. spreader it was able to handle the hot water flows with a reusable nozzle less O ring. The pop-out SS design with flush O ring seal provided a food-grade surface for the industry it served. Without penetrating the vessel until deployed it was ideal for spray dryers and cyclones.

Dan Guaricci is vice president of US operation, ATEX Explosion Protection LLC, Davenport, FL. For more information, call 863-424-3000 or visit www.atexus.com.

For other news, articles, and equipment reviews, visit our Explosion Protection & Safety Equipment Zone

Click here for Explosion Protection & Safety Equipment Manufacturers

Get information or register for Powder Show Toronto, June 4-6, 2019