Research into the sprinkler protection of automated storage systems used by IKEA has recently been undertaken, as Magnus Arvidson reports.

The Department of Fire Technology at SP Technical Research Institute of Sweden has recently conducted sprinkler tests inside a VASS (Vertical Automated Storage System) unit for IKEA Services AB. The objective of the tests was to determine whether sprinklers installed at the top of the unit are able to protect the integrity of the unit during a fire and thereby limit the risk for fire spread to adjacent objects or building parts.

Figure 1 The installation of the VASS unit inside SP’s fire test hall

The concept of VASS units
A VASS unit is an enclosed system of vertically arranged trays, an extraction platform and a series of computerised controls that delivers items to an ergonomically positioned workstation. The unit automatically locates and retrieves stored items with the push of a button or scan of a barcode. The trays handled by the system can be equipped with partitions and dividers, totes and containers, cases and cartons, etc. The benefits of this storage concept include maximum use of available space, increase productivity and improved worker safety.

The storage inside a unit is very compact. Technology is used to scan every tray’s height and the trays are automatically stored in the optimal space in the unit. A built-in weight scale automatically checks a tray’s weight to protect against overloading of shelves.

It should be emphasised that the probability for a fire inside a VASS unit is low. The primary sources for fire ignition are related to computer control and the electrical motors associated with the extraction platform. Unauthorised personnel do not have access to the unit. Despite the high level of intrinsic safety, a built-in fire protection system is often required.

The VASS unit used by IKEA
The VASS units used by IKEA are designed to fit their specific needs regarding the dimensions of the trays and the overall capacity. The trays are constructed out of steel, with solid side plates. The individual trays measure 4m x 0.8m (W x D). The rim height is 250mm and, in the tests, two dividers were installed such that three compartments were formed per tray. The unit used in the fire tests had at total of 45 trays, 25 trays on the back wall and 20 trays on the front wall.

In practice, the most common commodities stored in VASS units at IKEA are products, e.g. kitchen articles, doors, drawers and textiles, or similar items. Products from unexpanded plastics are not as common and products from expanded plastics are prohibited.

The unit used for the tests measured approximately 8m in height, 4.4m in length and 3m in depth. The unit was supplied by Kardex, Bellheimer Metallwerk GmbH and installed inside the fire test hall at SP by personnel from Kardex Scandinavia AB. The installation of the unit is shown in figure 1.

Objective of the tests
Prior to the tests, it was concluded that the use of solid trays (with solid side plates) will prevent water from sprinklers at the ceiling to reach the seat of the fire. Therefore, fire extinguishment or fire suppression by the activation of sprinklers installed at the top of the unit was not expected. Additionally, the trays are fitted with holes in the ends of each tray that will drain water to the sides. This feature is designed to prevent the top trays from being filled with water, jeopardizing the stability of the unit.

It was also understood that the enclosed construction of the VASS unit and the solid design of the trays will in itself delay the spread and intensity of a fire. The steel trays restrict the possibilities for vertical fire spread and the compact storage restricts air movement, which limits the oxygen concentration at the seat of the fire. When filled with combustion gases, the oxygen concentration inside the unit as a whole is reduced, which will reduce and possibly even quench a growing fire.

Therefore, it was recognised that the limitation of the fire size will primarily be an effect of a reduction of the oxygen concentration by the fire itself and to some extent by the formation of water vapour. In order to facilitate the desired reduction of oxygen, it was therefore deemed essential that the structural integrity of the VASS unit remains intact during a fire. As a result, the following criteria were set up as minimum requirements to judge the test results as acceptable:

• the VASS unit must remain structurally intact
• no full or partial rupture of any of the wall/ceiling panels or joints between these panels is allowed
• no rupture of the hatch for the front opening is allowed, and
• no fire spread to adjacent objects or combustible material is allowed.

The sprinkler system
The sprinkler system was intentionally designed to be relatively simple. The system pipework was laid out on top (outside) of the ceiling with drop nipples to the individual sprinklers. Four pendent, standard coverage, quick-response sprinklers were installed. The sprinklers were installed to ensure wetting of the inside surface of the walls of the unit.

An optical – photoelectrical – type smoke detector was installed at the ceiling of the unit. The response time of the detector was recorded in order to provide a comparison with the activation time of the sprinklers.

Instrumentation and measurements
Thermocouples were installed inside the unit in order to document the potential fire spread. Additionally, the surface temperatures on the outside of the unit were measured using thermocouples that were spot welded directly to the outer steel panels. A thermal imaging camera was positioned behind the back wall with the intention of documenting the thermal exposure in the area where the fire was initiated. The concentration of oxygen was measured at the centre line of the unit and 1.3m and 3.2m, respectively, above floor level. The lowest position (1.3m) was at the level of the tray where the fire was initiated.

Figure 2: The EUR Std plastic commodity that was used in the tray where the fire was initiated

The EUR Std Plastic commodity was used in the tray where the fire was started and in the trays directly above. This commodity is similar to the FM Global Standard Group A Plastic commodity, which has been widely used in the fire protection community to create a representative 'benchmark' warehouse fire hazard for the evaluation of sprinkler fire protection performance in large-scale fire tests since the 1970s. Authentic IKEA commodities (pillows of synthetic material and plastic crates) were used below and above these trays. All other trays were filled with packages of particle board or packages with kitchen doors from IKEA's product line. This commodity represented a realistic load and acted as an indicator of fire spread. Figure 2 shows the EUR Std Plastic commodity that was used in the tray where the fire was initiated.

Two tests
Two tests were conducted. In both tests, the fire was initiated in one of the bottom trays at the back wall of the unit. For the first test, the unit was loaded as would be expected under normal conditions, i.e. the density of the loading of the trays was high. The fire growth rate during the test was extraordinarily slow, due to the solid design of the steel trays that restricted the possibility for vertical fire spread. It is also clear that the compact storage (the minimum free vertical distance between individual trays was 25 mm) restricts air movement and limits the oxygen concentration at the seat of the fire. The fire detector at the ceiling responded at 7 min 15 sec but the first (and only) sprinkler did not activate until after more than an hour. The oxygen concentration inside the unit at the activation of the sprinkler was approximately 13.5 vol %. The oxygen concentration increased after the activation, probably due to mixing effects and a pressure reduction due to cooling by the water droplets, which drew fresh air into the unit. As the fire continued to burn, the oxygen concentration was further reduced and did not increase until manual firefighting was undertaken, 30 minutes after activation of the sprinkler.

The surface temperature at the wall panel closest to the fire was approximately 200°C at the activation of the sprinkler. The application of water, however, did not immediately reduce the surface temperature at this position. Rather, the temperature continued to increase and peaked at 405°C. The closest thermocouple, 1.1m above this point, was reduced from approximately 110°C to approximately 80°C by the application of water, which indicates that the high temperature exposure was very local. Figures 3 through 5 shows the thermal image of the back wall during the test.

Figure 3 Test 1. The thermal image immediately prior to activation of the first (and only) sprinkler. The fire has spread towards the back corner of the unit (to the right in the figure), as indicated by the increased temperatures	Figure 4 Test 1. The thermal image at approximately 63 min 13 sec, i.e. one minute after activation of the sprinkler. The flow of water down the inside of the wall is clearly seen	Figure 5 Test 1. The thermal image at approximately 72 min 13 sec, i.e. 10 minutes after activation of the sprinkler. The temperature exposure is very local

The thermal images may be compared with figure 6, where the burn marks can be seen on the back wall after the fire test.

Figure 6: The burn marks on the back wall after the first test, which indicate fire spread towards the back corner of the unit

For the second test, the tray directly above that where the fire was started, was removed. The free vertical distance between the fire tray and the tray above was therefore increased from 25 mm to 300 mm. As a result, the fire growth rate was significantly higher than that in the first test. This translated to a more rapid reduction of the oxygen concentration inside the unit. The fire detector at the ceiling responded at 2 min 49 sec and the first (and only) sprinkler activated at 8 min 53 sec after ignition, at an oxygen concentration of approximately 14.5 vol %. The oxygen concentration continued to fall, as the fire was not extinguished by the application of water, and reached a minimum level around 11.0 vol % at approximately 14 min 00 sec. This oxygen level was low enough to quench the fire. Manual firefighting was not necessary. The surface temperature at the wall panel closest to the fire reached 600°C. The application of water did not immediately reduce the surface temperature at this position. The closest thermocouple, 1.1 m above this point, peaked at 57°C which indicates that the high temperature exposure was very local. Figures 7 through 9 shows the thermal image of the back wall during the test. These figures show that the fire spread in both directions on the tray from the fire source and heated up the whole length of the wall panel.

Figure 7: Test 2. The thermal image immediately prior to activation of the first (and only) sprinkler. The fire spread in both directions in the tray and heated up the whole length of the wall panel	Figure 8: Test 2. The thermal image at approximately 9 min 53 sec, i.e. one minute after activation of the sprinkler. The flow of water down the inside of the wall is clearly seen	Figure 9: Test 2. The thermal image at approximately 18 min 53 sec, i.e. 10 minutes after activation of the sprinkler. The temperatures are very low across the full width of the wall

Conclusions
As expected, fire extinguishment from an installed sprinkler system may not be guaranteed in practice due to the shielded nature of the fire. However, automatic sprinklers installed at the ceiling of a VASS unit will maintain the structural integrity of the unit during a fire and the temperature exposure to the wall panels will remain local. It was also verified that that the probability for fire spread to adjacent combustible objects or material outside of the unit is very small, as only small parts of the wall were heated up.

As fire extinguishment is not guaranteed, it is advisable that a strategy, equipment and instructions for manual fire fighting are developed. The following strategy is suggested:

• the VASS unit should not be opened during a fire, use water from fire hoses to cool the structure from the outside
• use an infrared (IR) thermal imaging camera to help locate the fire
• drill (small) holes and insert lances at the seat of the fire, use water to fight the fire locally, and
• allow time for cooling before the unit is opened and ventilated.

A final word of caution is necessary. The experience described in this article is not directly applicable to other types of compact storage units. The geometry, construction and design of the units used by IKEA are unique and intended for their commodity and purposes. Other type of compact storage units that do not have the features of IKEA’s units may require other types of fire protection systems or other sprinkler system designs, in order to meet the desired fire protection objectives

Magnus Arvidson is a fire protection engineer at SP Technical Research Institute. He participated in the development of the first edition of NFPA 750 Standard for Water Mist Fire Protection Systems, and is a member of the CEN working group developing a similar standard for the European market. Should you have any further questions about your specific storage system please contact the author.