About Us

Preservation Unit 
425 Library, MC-522
UIUC Library
1408 West Gregory Dr.
Urbana, IL 61801

Conservation Lab 
Oak St. Library Facility
OSLF, 2nd Floor
809 South Oak Street
Mail Code 527
Champaign, IL 61820

Digital Content Creation 
413 Library, MC-522
UIUC Library
1408 West Gregory Dr.
Urbana, IL 61801

Sprinkler Test Simulation for HDS Shelving Array

Participants and contributors

On Monday, April 27, 2009, the UIUC Preservation Unit in collaboration with a Senior Engineering Team from the UIUC, IESE and in coordination with IFSI and AFST conducted a live sprinkler test on a 5 foot section of high density storage shelving.  The purpose of this test was threefold:

Development of testing protocol


Figure 1

The two storage modules in the Oak Street Library Facility are outfitted with 94 K-22 Early Suppression Fast Response (ESFR) Reliable sprinklers, each of which activates independently when temperatures exceed 200ºF.  The K-22 sprinkler head is designed to discharge 142 gallons of water per minute at 40 psi to prevent the fire plume early in its development, stopping the fire from spreading and reducing roof level temperatures quickly to prevent structural damage to the shelves and the facility.  This sprinkler head is designed to extinguish fire – not suppress fire spread.

Installation specifications for the K-22 ESFR are based upon an open rack/pallet system (where discharged water could actively Test configuration mock-uppenetrate within the shelving array, quickly extinguishing an active fire event).  To date, there has been no published testing of these types of sprinklers within a closed rack environment (like that found in the HDS facility).  Investigations looking at fire movement within a closed shelving system and the effectiveness of sprinklers (including the use of in-rack sprinkler systems) in fire management within these types of systems are currently being conducted at other partner institutions.  It should be noted, however, that the UIUC HDS does not have in-rack sprinklers, so our system will react differently from that being tested.

In light of the differences between closed and open rack shelving, the engineering team was interested in gauging the effectiveness of the K-22 ESFR system as currently installed and in developing a greater understanding of the flow and movement of the water upon the shelving towers.  Using a NIST Fire Dynamic Simulator (a computational fluid dynamics (CFD) model of fire-driven fluid flow) the team was able to model the water flow and spray pattern of the facility’s sprinkler heads by specifying parameters such as pressure, velocity, flow, and sprinkler head size.  The simulation showed that the top 5-7 feet of the shelves receive the majority of the water (see figure 1); spots below this are reached indirectly through cascading water.

The team observed that levels of pooling within the upper range of the shelves varied depending on the configuration of sprinkler head to shelf array (the sprinkler heads are positioned in a  9’ grid pattern to allow for overlap of the water cones in the event of a deployment – there is no intellectual correlation between the location of the sprinkler heads and the location of the shelving below,  as per the open rack specifications) and determined which configuration would result in a “worst case scenario” (i.e. the sprinkler head to shelving configuration that would result in the maximum amount of direct water penetration and damage) to be replicated within the live sprinkler test.

Materials and methods

Applying the information learned via the computer modeling, it was determined the test configuration would be as follows (see Figure 2):

Inter-test observations

Figure 8 & 9Figure 6 & 7The observable flow of water upon the array was, in general, as anticipated by the modeling software.  The trays lining the front of the upper shelves were directly hit by the sprinkler spray in a severe downward motion (causing both pooling of the water within the upper trays and shelves as well as a shearing of the front of the trays and books) and a cascade could be seen forming lower upon the array.  What the software and internal team discussions did not foresee was the ‘waterfall’ effect that was observed when the large amounts of high pressure water hit the top ‘roof’ of the shelving tower (see figures 6 & 7).  This ‘launching’ effect caused a continuous stream of water to fall into what would be the upper shelves of the back facing array.

The two special collections trays failed at 13 minutes into the test (the tray located in the far right facing position on the second shelf) and 16 minutes into the test (left facing, top shelf).  Failure of these trays appears to have been caused by an inherent weakness in the tray construction material (acid and lignin free fluted cardboard).  This can be assumed by the observable failure of the tray (the front was sheared off in a downward direction) prior to the loss of books from the tray (a two minute lag was observed between the failure of the tray front and the loss of books from the tray).  This would appear to indicate that the tray failed due to water penetration as opposed to book swell (see figures 8 & 9).

Post-test observations

After the 30 minute testing period was completed, the team was able to more fully assess the observable physical damage caused by the sprinkler deployment to the upper shelves of the array (see figures 11-14).

Post-test damage assessment









Post-test analysis

After the materials were removed from the shelves, the trays were reweighed and photographed.  The tagged books were then removed from the trays and measured, weighed (both in and out of their assigned housing, if appropriate), and photographed.

        chart 1

        chart 2

         chart 3

Recommendations and areas for future investigation

The testing confirmed that the top 5’ of shelving will experience an increased amount of damage, due to direct penetration, in the event of a sprinkler deployment.  In light of the observations and assessments made above, the following recommendations can be issued:

  1. Whenever possible, the upper shelves (especially the 1st through 3rd) of HDS ranges should be populated with those materials that would not be severely damaged by direct water penetration or those that, once wet, would need to be kept wet while awaiting conservation treatment.  These materials could include microfilm and microform formats as well as some types of digital media.
  2. Special collections should not be stored in the upper ranges of the HDS ladder.  
  3. Attention should be given to the materials shelved in the back row of the first and second highest shelves of the array.  These spaces will receive an increased amount of water and should be occupied, when possible, by materials that will not be inalterably damaged by prolonged water exposure.
  4. Labels and barcodes should either be duplicated (i.e. front and back of tray labeling) or be provided some form of additional protection from the water cascade.  Observed loss of both archival storage labels and HDS barcodes could pose challenges to intellectual control over the collection material.
  5. The failure of the special collections trays is a cause of great concern.  Attempts should be made to re-formulate either the design or the construction materials used in the manufacture of these trays to rectify this issue.  If it is not possible to create trays that will stand-up to prolonged water exposure, then discussions should be made to assess the storage priorities for these types of collections.
  6. While the archival record storage box and folders performed well as protective barriers against direct water penetration, the box construction (i.e. use of water-soluble paper tape) was not able to be easily removed from the shelf without disturbing the contents.  Record storage boxes that are to be housed in the HDS facility should be constructed with side staples and not paper tape (in order to facilitate movement from the shelf to the area of triage/assessment.


Follow-up testing: special collections housing and triage test

Participants and contributors

On Monday, August 24, 2009, the UIUC Preservation Unit conducted a second live sprinkler test on a 5 foot section of high density storage shelving.  The general goals of this test were to:

  1. Assess a beta special collection tray manufactured by Metal Edge for the purpose of this test.  In response to the failed trays recorded in the April test, Metal Edge produced a series of acrylic coated special collection trays with the intention being to expose the materials to less water penetration.  
  2. Observe the protective qualities of housings frequently utilized by the Special Collections units including:
    • Record storage boxes with paper tape side seams;
    • Metal Edge document storage boxes;
    • Rolled storage tubes.
  3. Observe the potential level of water penetration into a flat file storage cabinet.  
  4. Implement, to the extent possible, the proposed disaster recovery protocol - including, but not limited to, the triaging of materials unsuitable for stabilization through freezing.

Methods and materials

Figure 17Second test setupThe test followed the methods and materials utilized in the first sprinkler deployment test with the following modifications:

figure 18Inter-test observations

The first tray failure (first shelf, first position – acrylic coated test tray) was observed at 5:33 minutes into the 30 minute test.  As was observed in the first test, tray failure appeared to be due to an inherent mechanical weakness in the tray material (as opposed to bookswell) as the first book did not fall from the tray until 6:49 minutes into the test (over a minute later).  The second tray failure occurred at 8:07 (second shelf, third position – coated test tray).

The third failed tray was observed at 9:16 as one of the back shelf trays gave to book swell (these back trays were experiencing the waterfall that formed when the high pressure water was pushed off of the top most ‘roof’ shelf of the front array (see figure 18). 

The regular special collections trays (i.e. uncoated) failed at the 12:39 mark (fourth shelf, fourth position) and at 16:29 (third shelf, second position).  The timing of the two uncoated special collection failures was similar to those observed during the first sprinkler deployment test. 

The final tray failure (back shelf) occurred at 18:12, again due to book swell.


Post-test observations

Figures 19 & 20After the 30 minute testing period was completed, the team was able assess the observable physical damage caused by the sprinkler deployment.


Post-test analysis

         chart 4

         Chart 5

Recommendations and areas for future investigation

  1. The application of an acrylic coating to the special collection tray construction materials did not increase the stability of the trays.  While the acid-free/lignin-free qualities of the trays are ideal for the long-term storage of rare books and special collections, the fact that in the event of a water disaster these trays will most certainly fail is a cause of concern.  Due to the high quality of the environment found in the HDS facility, the negative impact of the general collection trays may not be as detrimental as would be expected in a less climate controlled environment.  Curatorial and preservation staff should reconsider the use of general collection trays for the storage of all special collection materials.
  2. The unique coloring of the special collection trays has been a factor in facilitating the identification of some special collection materials.  If all collection materials are to be shelved in ‘general collection’ trays, a mechanism will need to be put in place to assist in the identification of special from general collection materials.  Investigations should be made into the application of colored label savers.  Use of tinted label savers would serve the dual function of identifying special collection trays and protecting barcodes (and archive labels) from water damage in the event of a sprinkler deployment.
  3. The failure of the paper-tape side seams on the tested archival record storage boxes was an additional cause for concern.  The tape failure completely undermined the structural stability of the box construction and made the safe extraction and triage of these materials excessively difficult.  All record storage boxes that are sent to the Oak Street Facility for storage should have side-staple seams (not paper-tape).  
  4. The quick identification of ‘problematic’ materials stored in record and Metal Edge document storage boxes is will be a high priority for triage staff in the event of a large scale water disaster.  A sticker/label that identifies that there is (or is not) AV, photographic, and/or data storage media present in the box should be developed with the special collections curators for use on ALL closed boxes.  
  5. While the flat file portion of the test was inconclusive, the need for a large cart with which the safe retrieval of the flat file materials could be accomplished was noted.  Additionally, a discussion should be instigated with BMS to determine how they would handle the transport and storage of oversized flat files in the event of a disaster at the HDS facility.


Tabular data

Table 1: First test – tray data
Table 2: First test – Tagged book data
Table 3: Special Collections Housing test