HVLS Fans and Sprinkler Operation Phase 1 Research Program

Transcription

1 HVLS Fans and Sprinkler Operation Phase 1 Research Program Final Report Prepared by: Schirmer Engineering Corporation \ The Fire Protection Research Foundation One Batterymarch Park Quincy, MA, USA foundation@nfpa.org Copyright Fire Protection Research Foundation February 2009

2 FOREWORD High volume low speed (HVLS) fans are in increasing use in storage and manufacturing facilities. However, the interaction of these fans and automatic sprinkler operation is unknown. In order to inform spacing and other installation requirements in NFPA 13, Standard for the Installation of Automatic Sprinklers, the Foundation initiated a research program. Phase I of this program which includes a literature review and small and large scale testing with a focus on ESFR sprinklers and rack storage is described in this report. The content, opinions and conclusions contained in this report are solely those of the author.

3 HVLS Fans and Sprinkler Operation Phase 1 Research Program PROJECT TECHNICAL PANEL Craig Beyler, Hughes Associates Inc. Prateep Chatterjee, FM Global Ken Isman, National Fire Sprinkler Association Jim Milke, University of Maryland Joe Noble, Noble Consulting, IFMA representative Don Norris, Cintas Fire Protection Jeff Tubbs, Arup Fire Jim Lake, NFPA staff liaison PRINCIPAL SPONSORS XL Global Asset Protection Services Big Ass Fans CNA Liberty Mutual Macro-Air Technologies Zurich NA CONTRIBUTING SPONSORS Ace American Insurance Company Allianz Risk Consultants Rite Hite Corporation

4 HVLS FAN AND SPRINKLER OPERATION PHASE I RESEARCH PROGRAM FINAL REPORT THE FIRE PROTECTION RESEARCH FOUNDATION ONE BATTERYMARCH PARK QUINCY, MASSACHUSETTS Prepared For: Ms. Kathleen Almand Executive Director The Fire Protection Research Foundation One Batterymarch Park Quincy, Massachusetts SEC Project No.: February 17, Schirmer Engineering Corporation This report is the property of Schirmer Engineering Corporation. Copies retained by the client shall be utilized only for his use and occupancy of the project, not for the purpose of construction of any other projects.

5 Fire Protection Research Foundation i February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: TABLE OF CONTENTS EXECUTIVE SUMMARY... iii INTRODUCTION/PROBLEM STATEMENT...1 LITERATURE REVIEW...1 EXPERIMENTAL TEST PLAN...5 INTRODUCTION OF CONCEPTUAL PARAMETERS...5 ACTUAL DELIVERY DENSITY (ADD) TESTING...8 FULL-SCALE FIRE TESTING...10 EXPERIMENTAL RESULTS AND ANALYSIS...13 ACTUAL DELIVERY DENSITY TESTING...14 FULL-SCALE FIRE TESTING...21 CONCLUSION AND RECOMMENDATIONS...25 RECOMMENDATIONS FOR PHASE II RESEARCH...29 BIBLIOGRAPHY...34 INDEX OF TABLES TABLE 1. UL 1767 TABLE 30.1 ADD TEST PARAMETER SUMMARY [10]...9 TABLE 2. PHASE I ADD TEST PLAN...10 TABLE 2 (RE-PRINTED). PHASE I ADD TEST PLAN...14 TABLE 3. STANDARD PRESENTATION OF ADD RESULTS (EXTREME CASES)...16 TABLE 4. COMPARISON OF ADD RESULTS FOR 2-SPRINKLER CONFIGURATION...18 TABLE 5. COMPARISON OF ADD RESULTS FOR CHANGE IN VERTICAL CLEARANCE...20 TABLE 6. PROPOSED AIRFLOW EXPERIMENTATION...30

6 Fire Protection Research Foundation ii February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: TABLE OF FIGURES FIGURE 1. HVLS FAN PRODUCT IN A COMMERCIAL SETTING [1]...1 FIGURE 2. CENTRAL SOLIDITY RATIO CONCEPT...6 FIGURE 3. CENTRAL SOLIDITY RATIO COMPARISON...6 FIGURE 4. FLOATING SOLIDITY RATIO CONCEPT...7 FIGURE 5. UL 1767 ADD TEST DIAGRAMS [10]...9 FIGURE 6. STAGNATION FLOW PATTERN...11 FIGURE 7. WORST-CASE FAN LOCATION FOR UL/GE GAP TEST SERIES [8]...12 FIGURE 8. DIAGRAM OF FIRST FULL-SCALE FIRE TEST AT UL...13 FIGURE 9. SAGGING OF SUPPORTING STEEL TRUSS...15 FIGURE 10. OBSERVATION OF PAN WARPING...15 FIGURE 11. EXPERIMENTAL OBSERVATION OF FLOW SHADOWING...17 FIGURE 14. OBSERVATIONS OF FIRE SPREAD IN FULL-SCALE TEST FIGURE 15. UL DIAGRAM OF SPRINKLER ACTIVATIONS FULL-SCALE TEST FIGURE 16. EARLY FIRE GROWTH WITHIN LOWER TIERS TEST FIGURE 17. UL DIAGRAM OF FULL-SCALE TEST 2 SETUP...25 FIGURE 18. LOGIC TREE FOR PHASE II FULL-SCALE FIRE TEST SERIES...32 APPENDIX APPENDIX A SOUTHWEST RESEARCH INSTITUTE FIRE TEST REPORT ADD TESTING APPENDIX B UNDERWRITERS LABORATORIES FIRE TEST REPORT FULL-SCALE TESTING APPENDIX C REQUEST FOR PROPOSALS PUBLISHED BY FPRF APPENDIX D COST OPINION FOR PHASE II RECOMMENDATIONS

7 Fire Protection Research Foundation iii February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: EXECUTIVE SUMMARY The objective of the Phase I research effort was to explore the interaction between a High Volume Low Speed (HVLS) fan and Early Suppression Fast Response (ESFR) sprinkler protection of rack and palletized commodity storage. This interaction was specifically defined in terms of the obstruction of sprinkler discharge and the required fan shut-off time, including detection means. The obstruction of sprinkler discharge was investigated in a series of Actual Delivery Density (ADD) tests designed in accordance with UL These tests were performed to evaluate the hypothesis that obstruction severity would increase with increasing size and proximity of the static fan obstruction to the sprinkler. The hypothesis was neither proven nor disproven by the test method, which provided very little information regarding the role of the obstruction in manipulating droplet trajectories. Additionally, droplet sizes and velocities are known to be important characteristics of the distribution pattern. No quantifiable information regarding these parameters can be provided by standard ADD testing. Literature review revealed that full obstruction of a single ESFR sprinkler (plugged sprinkler) is not a failure point for the system. With this in mind, the search for a worst-case fan placement relative to sprinklers focused on the obstruction of multiple devices. The results of the ADD tests indicated significant interference of the HVLS fan with sprinkler overlapping in a 2-sprinkler configuration. It is known that such overlapping, particularly in the first ring of sprinklers, is important to the success of the ESFR system. Therefore, fan placement between four sprinklers was pursued in subsequent full-scale fire testing. Additionally, this fan placement was selected to simplify the obstruction problem and was considered by the fan manufacturers as a commonly used and practical fan placement. A total of two full-scale tests were conducted at Underwriters Laboratories following the completion of the ADD effort. Both tests were performed using ESFR protection of rack storage of Group-A plastics stored to a height of 20-feet underneath a 30-foot ceiling. The first major objective was to verify the existence of a problem with an HVLS fan operating at maximum thrust and positioned to achieve the worst-case obstruction of sprinkler discharge. Fan placement was varied between tests with the worst-case results measured for fan placement centered over the point of ignition. The results apply only to this specific storage configuration as different boundary conditions will likely result in a different influence of the airflow on fire spread. It is thought that these results were more severe than the case of ignition below the tip of the blade because the former strategy promotes shielded fire growth in the space between storage tiers. Evaluation of the full-scale test results may be made in terms of the established pass/fail criteria as well as comparison to a historical baseline test conducted as part of the National Quick Response Sprinkler Research Project. Results of the two full-scale fire tests revealed no threat to the structure with ceiling gas temperatures briefly peaking at values on the order of 250 o F for both tests. A total of 8 sprinklers activated in the test in which the fan was positioned directly above the point of ignition (Test 2). This is significantly higher than either the first full-scale test (Test 1) or the baseline case, in which fire suppression was achieved with the activation of 2 and 3 sprinklers, respectively.

8 Fire Protection Research Foundation iv February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The increased number of sprinklers which activated in Test 2 may be attributed to the sequence of sprinkler operation as second-ring sprinklers operated prior to the four first-ring sprinklers. The National Quick Response Sprinkler Research Project report suggests activation of 8-10 sprinklers is not uncommon in that circumstance. Nevertheless, the total number of sprinklers activated during the second full-scale test was definitively less than 12, which is an established design criterion for ESFR systems. Activation of the first sprinkler was consistently at or very near one minute after ignition in all three cases. Transient ceiling temperatures located directly over ignition often mimic the rate of fire growth and therefore may be used to assess performance of the ESFR system. The convective currents introduced by the presence of an operating HVLS fan complicates any efforts to draw distinct correlations between these parameters; however, evaluation of these temperatures over extended time-scales still provides useful indicators of general fire growth or decay. In Test 1 these temperatures are suppressed after the activation of the first two sprinklers, implying suppression at the seat of the fire directly below. The response of these same gas temperatures in Test 2 was noticeably different. In this case, a shallow decline after the nearly simultaneous activation of the first three sprinklers is followed by slight re-growth. A gradual decay ultimately results as additional sprinklers activate. Water from sprinklers generally serves to absorb heat from gaseous flames and solid surfaces as well as limiting further fire spread by wetting unburned commodity surrounding the ignition site. The more gradual decline in ceiling gas temperatures in Test 2 suggests that this process occurred in a less efficient manner than Test 1. The timing of the sharp decline in ceiling temperatures in Test 1 is consistent with the definition of fire suppression given in NFPA 13. In contrast, the results of Test 2 appear to be more consistent with the definition of fire control given by the standard. As previously discussed, such performance from ESFR sprinklers is considered acceptable for the challenges presented in Tests 1 and 2. The authors then conclude that the presence of the HVLS fans did not affect the performance of the ESFR sprinklers to a level that would be considered unacceptable and that the performance of the sprinkler system is consistent with the criteria established in the baseline testing project. As a result of the Phase I research experiments, measurements and observations discussed in this report, the following conclusions and recommendations are made regarding obstruction of sprinkler discharge and the issue of fan shutoff: 1. Phase I results apply only to 30-foot high facilities with 20-foot high storage, which encompasses a significant portion of current HVLS storage and manufacturing applications. 2. The minimum vertical clearance between the fan obstruction and a sprinkler deflector at ceiling level should be 3 feet as currently allowed by NFPA 13 for clearance to storage. The effects of reduced vertical clearances of the obstruction (below 3 feet) were not explored in this research effort. 3. HVLS fans were consistently placed between 4 sprinklers in all full-scale tests conducted in Phase I. In an effort to build on the results presented in this report, it is recommended that HVLS fans be installed between 4 sprinklers. ADD testing was unable to definitively determine a worst-case fan placement relative to nearby sprinklers. Particle Image Velocimetry (PIV) and Phase Doppler Interferometry (PDI) techniques potentially offer a

9 Fire Protection Research Foundation v February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: means for obtaining definitive obstruction severity data, though these approaches are currently not standardized and also require a significant investment of resources. 4. The influence of fan-induced airflow on fire spread is a strong function of boundary conditions such as fan airflow (376,804 cfm used in Phase I) storage geometry, clearance from fan to storage, height of the test facility, distance to walls and the location of additional fans. In the configuration studied in Phase I, there is evidence that the storage arrangement significantly buffers the fan-induced airflow and minimizes its influence on fire spread. Based on this evidence, it was assumed that the worst-case placement of the fan relative to ignition in this configuration was directly above the latter. Further research is recommended to verify that this is indeed the worst-case placement of the fan relative to ignition. Such research could be conducted by a number of means including computational modeling [obstacles to using Fire Dynamics Simulator (FDS) noted in this report], scale modeling or additional full-scale testing. 5. The fan model chosen for testing in the Phase I research effort possesses a central solidity ratio of 0.60 as illustrated in Figure 3 of this report. According to this characterization, 83% of the fan models analyzed in this report possess nearly identical or less invasive shapes than the model tested in Phase I. 6. Successful system performance was achieved without fan shutoff in both full-scale tests. An attempt was made in Phase I to evaluate ESFR system performance in full-scale fire testing under practical worst-case fire conditions. The basis for generating these conditions relied heavily on the positioning of the fan relative to sprinklers and ignition location. It is clear that the flow field is strongly influenced by boundary conditions such as the geometry of the storage array, the clearance between the fan and storage, clearance from fan to room boundaries (i.e., walls and floors) and spacing between multiple operating fans where applicable. It is therefore recommended that a thorough analysis of the relevant physics be performed to determine characteristic fluid flow parameters responsible for enhancing fire spread (i.e., turbulence intensity). Such parameters could then be explored in non-fire airflow testing and confirmed in full-scale fire testing. The objective of this effort would be to either confirm or challenge fan placement relative to the location of ignition in the Phase I tests. To that end, a suggested test plan for Phase II of this program is provided within this report.

10 Fire Protection Research Foundation 1 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: INTRODUCTION/PROBLEM STATEMENT High Volume Low Speed (HVLS) fan technology was invented by Walter Boyd in 1995 as a lowmaintenance and energy efficient alternative to smaller fans operating at higher speeds. The principle behind this efficiency is the use of large airfoil type propeller blades influencing a correspondingly large volume of air as they move through their range of motion. By increasing the range of motion, design airflows may be achieved at lower blade speeds. The relatively light weight of the hollow foil extruded aluminum blades allows the fans to be operated by motors with 1 horsepower or less capacity, while operating at fewer rotations per minute (Figure 1). FIGURE 1. HVLS FAN PRODUCT IN A COMMERCIAL SETTING [1] The invention was conceived as a means to optimize the productivity of dairy cattle, though the advantages of maintaining a well-ventilated work environment for humans were soon realized. Recently, HVLS fans have become increasingly popular in storage and manufacturing facilities. However, such widespread usage preceded verified and validated engineering hypotheses on the potential for interference with the performance of ceiling level sprinklers. Consequently, specific design requirements remained absent from the National Fire Protection Association Standard 13 (NFPA 13) Standard for the Installation of Sprinkler Systems. The potential impediment to achieving fire suppression was brought to the attention of the Fire Protection Research Foundation (FPRF) by the insurance industry in conjunction with the NFPA 13 technical committee. Concerns focused on the degree of obstruction/alteration of sprinkler water spray patterns as well as the impact of fan induced airflow on the rate of fire spread through commodity arrangements. Consequently, a Phase I research program was designed by the FPRF with objectives focused on obstruction of sprinkler discharge and required fan shut-off time including detection means. Investigation of these objectives in Phase I was limited to ESFR sprinkler performance for rack and palletized storage of commodities. The scope of services included a literature review of fan performance and historical benchmark ESFR sprinkler tests, finalization of a detailed experimental test plan, coordination of the test plan with fire test laboratories, analysis and reporting of the results and recommendations for the path forward (Phase II). Schirmer Engineering was retained by the FPRF to manage the Phase I research program as the result of a public request for proposals process (Appendix C). The following document reports the findings of Schirmer Engineering generated by the efforts outlined in the scope of services.

11 Fire Protection Research Foundation 2 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: LITERATURE REVIEW The initial focus of the review is on the obstruction of water distribution from an ESFR sprinkler. Existing knowledge on this topic emphasizes the significance of the linear distance between the obstruction and the source (sprinkler). Generally, the concern over the severity of the obstruction increases as the object is moved closer to the sprinkler. Historically, much attention has been devoted to the effects of obstructions at short horizontal distances from the source and very tight to the ceiling as well as those obstructions positioned within the first 18 inches below the sprinkler deflector [2]. These areas constitute the volume of space in which the sprinkler discharge pattern evolves into a fully developed distribution. Obstructions in these critical near-field regions are potentially very disruptive to the intended distribution. In NFPA 13, rules for sprinkler spacing from obstructions in the near-field are strongly governed by sprinkler type and obstruction size (quantified by width). Far-field obstructions are generally disregarded. According to specifications obtained from four (4) major manufacturers, the desired installation location for HVLS fans is typically no less than 3 feet below an ESFR sprinkler deflector installed at ceiling level (with the exception of a thin ceiling-attachment piece). FM Global Data Sheet 2-2 characterizes the significance of obstructions located at this vertical distance below the sprinkler deflector in the following statement [3]: Obstructions located more than 36 in. (914 mm) below suppression mode sprinklers will not disrupt the discharge pattern, but cannot be ignored because they can obstruct water penetration into flues in storage. Obstructions that are located directly over flues must be at least 36 in. (914 mm) above the flue. Additionally, both FM Global Data Sheet 2-2 and NFPA 13 require 36 inches of clearance to storage when ESFR sprinklers are installed at ceiling level. Discussions with the scientists responsible for the development of the ESFR sprinkler indicate that the reason for this restriction is that at clearances less than 36 inches, water from discharging sprinklers can ricochet back upward onto adjacent sprinklers causing cold soldering of these sprinklers. The combined NFPA 13 and FM Global regulations suggest that when HVLS fan blades are located more than 36 inches below an ESFR sprinkler deflector and at least 36 inches above flue spaces, the obstruction posed by a static (not moving) fan is benign. The above conclusion is implied based solely on proximity without regard for obstruction shape or size. The general shape of an HVLS fan is uniquely complex. This complexity could be significant because knowledge in this area is generally extrapolated from experience with more simple obstruction geometries such as a single pipe or joist. Such obstructions are essentially two-dimensional with one of the dimensions being an infinite length with respect to the spray pattern. The width or diameter of the object is convenient for quantification of size. When the geometry of the obstruction becomes more complex, NFPA 13 currently recognizes solidity as a relevant parameter. Perhaps the best example is found in a provision known as the Three Times Rule (modified for application to ESFR sprinklers), which is intended to get water through (over and under) a relatively open obstruction such as an open web truss while avoiding any significant dry spots. Obviously this practice is ineffective for solid continuous obstructions that are impervious to water droplets. The point is simply that the solidity of an obstruction plays an established role (albeit one that is not well-understood) in determining appropriate sprinkler spacing.

12 Fire Protection Research Foundation 3 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: This investigation seeks to expand the base of knowledge on obstructions to ESFR sprinklers as it applies to HVLS fans used in storage and manufacturing applications. However, there is currently no verified and validated test method for evaluating sprinkler spray obstructions other than the execution of a parametric analysis in full-scale fire test mode. Nonetheless, due to limited resources, a parametric analysis of obstruction severity was conducted at an intermediate scale. The type of testing utilized is known as Actual Delivery Density (ADD) testing, which is a standard method of ESFR performance evaluation employed by both Underwriters Laboratories and FM Global. It is important to note that neither entity utilizes the test method to evaluate the effects of sprinkler spray obstructions on sprinkler performance. The lone support in the literature for this approach to evaluating spray obstructions is in the form of speculation asserting that useful information could potentially be obtained for investigating obstruction clearances [5]. The results of such an effort remain absent from the literature. Fortunately, at least one prominent source with limited information on ESFR sprinkler spray obstructions gathered by full-scale fire testing is available. The National Quick Response Sprinkler Research Project was conducted by the FPRF in an effort to evaluate a series of potential failure modes for K-14 ESFR sprinklers protecting double-row rack storage of plastics with 4-foot wide aisles. The only relevant spray obstruction data in this series was yielded in the results of a low-clearance test (approximately 3 feet from deflector to storage) with ignition located between 2 sprinklers, one of which was fully obstructed (plugged). The fire proved to be the most challenging of the test series with a total of 11 sprinklers activating. However, the result was declared a success by the authors of the study. This conclusion implies that ESFR sprinkler spray obstructions must influence multiple sprinklers in order to cause system failure [6]. The way in which system failure is defined may be a topic for debate; however, it is important to note that the current version of NFPA 13 allows designers to utilize criteria established by this specific literature source. Additional sources were reviewed to form a foundation for the Phase I research effort. These resources most notably included the following: Chicarello, Troup and Dean. The National Quick Response Sprinkler Research Project: Large Scale Fire Test Evaluation of ESFR Automatic Sprinklers, Fire Protection Research Foundation Report, May [2] Yao, Cheng. The Development of the ESFR Sprinkler System. Fire Safety Journal, 14 (1988) [3] Willse, P. and Pabich, M. Fire Test Evaluation on the Effects of High Volume Low Speed Fans on Ceiling Sprinklers, Global Asset Protection Services Research: Internal Publication and Presentation at 2008 NFPA World Safety Conference & Exposition. [4] Valentine, V. and Isman, K. Interaction of Residential Sprinklers, Ceiling Fans and Similar Obstructions. Fire Protection Research Foundation Report, January [5] Luna, M. and Trevino, J. Report of Special Project Testing of Big Ass Fan s Powerfoil TM 20-ft Ultra Fan Assembly to its affect on the Actual Delivered Density of a K = 14 Sprinkler Head to a modified UL 1767 Test, Intertek Report Number , May 29, [6]

13 Fire Protection Research Foundation 4 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The last of these sources was compiled as a report on fire testing by Intertek Testing Services in conjunction with Big Ass Fans, a major manufacturer. According to the test report, the original issue date to Big Ass Fans was May 29, However, the report was not made available to the Phase I research program, until approximately 3 months later at a time when the Phase I research was well underway. This factor limited the utility of the report in Phase I. The degree of relevance of each of the referenced experimental efforts to the stated objectives of the Phase I research program is varied. The first two of the listed sources deal exclusively with the development of the ESFR sprinkler with no specific consideration of HVLS fan equipment. The third source evaluates the impact of HVLS fans on the performance of sprinklers, but the type of sprinkler used (control mode) was characteristically different than the ESFR sprinkler. The fourth literature source does not include HVLS fans or ESFR sprinklers, but rather deals with a similar issue of spray obstruction by fan blades at the residential scale. Despite these varying degrees of relevance to the task at hand, a number of general conclusions drawn from these sources were used in forming the logic for the Phase I approach. These important pieces of general information were as follows: Control mode sprinklers, with characteristically low sensitivity and low rates of water delivery, failed to control full-scale fire tests with HVLS fans operating over high hazard storage configurations. ESFR sprinklers are much more effective due to higher sensitivity coupled with higher rates of water delivery. These different protection schemes are often utilized to meet different objectives ranging from fire suppression at one extreme to varying degrees of fire control at the other. There is an optimal operating sequence for ESFR sprinklers, which was observed in the National Quick Response Sprinkler Project. When this sequence is followed, fire suppression is anticipated by the first ring of sprinklers. However, successful system performance may also be achieved when the system does not respond in an optimal sequence. In such cases, as many as eleven sprinklers (multiple-rings) may be necessary to yield fire suppression. Based on the results of the National Quick Response Sprinkler Research Project, a single fully-obstructed sprinkler in the vicinity of the ignition area is not a failure mode for an ESFR system. This implies that the set of obstruction-induced failure scenarios includes only cases in which multiple sprinklers are obstructed. In general, obstructions to sprinkler discharge can prevent fire suppression by reducing the actual delivery density of water to the burning fuel, through reduction of droplet velocity and trajectory. This may be sufficient to cause sprinkler skipping. In tests with residential ceiling fans operating in the presence of residential sprinklers, when the fans were run at high speeds, noticeably slower sprinkler activation times were observed. This was attributed to the fan-induced dispersion of convective heat produced by the fire. The overall impression from these general conclusions in the literature was that the combined spray obstruction and dispersion of convective heat from the fire plume produce significant challenges for residential and control mode sprinklers. No definitive data was found regarding the effects of HVLS fans on the performance of ESFR sprinklers.

14 Fire Protection Research Foundation 5 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: EXPERIMENTAL TEST PLAN The stated objectives of the Phase I research effort were to investigate the obstruction severity of sprinkler discharge patterns as well as to determine a method and timeline for fan shut-off if necessary. Based on the structure of these objectives, the experimental approach was divided into two distinct building blocks. The first of these sub-phases was aimed at understanding the role of various parameters in determining the obstruction severity of a static (shut-off) fan. The intent of this approach was to assume that an acceptable method and timeline for fan shut-off could be achieved and consequently to determine if the static fan obstruction alone was significant enough to defeat the ESFR system. Introduction of Conceptual Parameters The experimental investigation therefore began with a simple hypothesis that the severity of the discharge obstruction would increase with increasing size and proximity of the obstruction to the source of sprinkler discharge. Some complexity was introduced to this simple hypothesis by the extent of HVLS fan design parameters dictating variances in the characteristic obstruction posed by different fan models and manufacturers. Consequently, an effort was made to condense these geometric parameters into a single expression for obstruction severity with application to a wide population of HVLS fan products. The form of expression chosen for this comparison was a solidity ratio as defined in Equation 1 and illustrated in Figure 2. Solidity Ratio solid = (1) A solid A + A open Where: Asolid A open = Total solid-object area within circular region of interest = Total open area within circular region of interest As will be shown later in this report, HVLS fans present a unique type of obstruction in which solidity currently varies from approximately 30% to 70% as calculated in Equation 1. The parameter largely responsible for this variation in solidity is the blade width, which is uniform to all blades of the same fan. Yet simply characterizing the obstruction in terms of blade width seems inadequate for cases when the solidity ratio is high and the obstruction is perhaps more like a continuous large-diameter disc. The solidity ratio is therefore introduced as an attempt to paint a more complete picture of the object than the blade width can provide. Another critical point to be made is that the fan diameter is much larger than the diameter of the sprinkler spray. For this reason, the solidity ratio focuses on a roughly 6-foot diameter region. This corresponds to the approximate diameter of the sprinkler spray, 3-5 feet vertically below the sprinkler, where HVLS fans are typically mounted in this application. The total solid area within this region is maximized when the fan and spray centerlines coincide. When the fan is moved away horizontally the total solid area within the fixed region decreases. The former case is therefore used as a basis for comparison of the maximum-possible solidity ratio or central solidity ratio posed by a number of HVLS fan models on the market today (Figure 3). The blade width of the selected sample of fan models in Figure 3 varies from about 7 9 inches.

15 Fire Protection Research Foundation 6 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 2. CENTRAL SOLIDITY RATIO CONCEPT Approximate spray boundaries FIGURE 3. CENTRAL SOLIDITY RATIO COMPARISON MacroAir (Wickerbill, Whisperfoil, Whisperfoil XL) Big Ass Fans (Powerfoil) Envira North (Stealth, Stellar, Zephyr) Rite Hite (Revolution) 0.70 Central Solidity Ratio Fan DIameter [feet]

16 Fire Protection Research Foundation 7 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The plot illustrates designs posing a worst-case obstruction severity of anywhere from 30%-70% of the sprinkler spray as previously indicated. Without the aid of experimental data, it is unclear whether such solidity could prevent sprinkler discharge from reaching the hazard. It is anticipated that high solidity values will require sprinkler spacing for obstruction avoidance, similar to current practices for avoiding continuous obstructions near the ceiling. The definition of what qualifies as a high solidity ratio is therefore sought. Based on the data in Figure 3, the worst-case scenario would be an obstruction with 70% solidity. However, limited testing resources dictated the use of an approximately 60% central solidity ratio product. This product was ultimately supplied by MacroAir, one of the sponsors of the Phase I research program. The exact model was a 24-foot diameter, 10-blade MaxAir Whisperfoil XL with a central solidity ratio of The previously stated experimental hypothesis was that spray obstruction severity is a function of both size and proximity to the sprinkler spray. The concept of the central solidity ratio allows for a quantifiable expression of obstruction size, but the issue of proximity introduces a variation on this concept. The reason for this variation is that in many potential cases, the boundaries of the sprinkler spray do not correspond to the central region of the fan. Therefore, it is necessary to consider a floating solidity ratio as illustrated in Figure 4. In a situation where the sprinkler spray is centered directly over the fan hub, the floating solidity ratio is the central solidity ratio. The significance of the floating solidity ratio concept is that it illustrates how a single fan can obstruct different percentages of a fixed spray region. This is accomplished simply by moving the fan around in the spray and determining a suitable method of measuring the effects. In other words, simply moving the 24-foot diameter, 10-blade MaxAir Whisperfoil XL through different locations in the fixed spray can produce floating solidity ratios anywhere from 0.00 (far away) to 0.60 (centered on sprinkler). This is a good range considering that nearly 83% of the fans shown in Figure 3 are incapable of yielding larger solidity ratios regardless of position. One factor not accounted for by the solidity ratio is the fact that the obstruction shape changes as the fan floats around within the fixed spray. This plays a role in the interpretation of test results shown later in this report. FIGURE 4. FLOATING SOLIDITY RATIO CONCEPT Approximate spray boundaries

17 Fire Protection Research Foundation 8 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: Actual Delivery Density (ADD) Testing For a fundamental approach to the complex physics of fire suppression by water, it is reasonable to assume that the portion of water droplets finding their mark are either evaporated in the flame or on the burning fuel surface [8]. The former retards the rate of thermal feedback to the fuel surface while the latter attacks the source directly. At any point in fire development, a certain critical amount of water is necessary to yield suppression. This idea applies broadly to methods of water based fire suppression, though it has been specifically standardized for use in characterizing ESFR sprinklers in terms of Actual versus Required Delivery Densities. The most important parameters characterizing the efficiency of a water distribution pattern are the size, velocity and trajectory of the droplets. Each of these parameters may be significantly influenced by the presence of an obstruction, yet to date there is no tried and true method of systematically evaluating such effects. Some momentum has been generated in recent years by the complimentary use of particle image velocimetry (PIV) and phase Doppler interferometry (PDI) to measure mean drop velocity and size in a non-intrusive manner [9]. Although such an approach to evaluating the spray obstruction posed by HVLS fans may produce technically superior information, the necessary investment of resources to collect and analyze such data at this stage was prohibitive. A more rudimentary approach to evaluating water distribution patterns from automatic sprinklers was conducted by Beyler [10]. A total of ninety-four experiments were conducted in a parametric analysis of the influence of design variables such as supply pipe size on the sprinkler spray distribution collected in floor level containers during a non-fire situation. This effort produced several insightful conclusions including an observance of what was termed the shadow effect in reference to low density distribution directly below a supply pipe for upright sprinklers. This type of approach to evaluating HVLS obstructions was deemed feasible with available project resources and also of some benefit in assessing relative severity of certain configurations despite the lack of information provided on the critical parameters of droplet size and velocity. The use of a similar standardized test was preferred for reasons primarily associated with reproducibility. Standards for the evaluation of ESFR sprinkler performance have been published and maintained by both FM Global and Underwriters Laboratories. Both standards include ADD testing with minor differences in the method of approach. For the purposes of this Phase I research effort, the 2005 edition of UL 1767 Early Suppression Fast Response Sprinklers was analyzed. This standard ADD test consists of positioning a heptane fire with a constant gas supply beneath configurations of 1, 2 or 4 operating ESFR sprinklers. Water collection pans are positioned beneath the flames in an effort to collect the spatial distribution of water droplets that pass through the flame without evaporating (Figure 5). The standard ADD test offers a bench-scale method of assessing the relative significance of parameters such as on the fundamental sprinkler spray distribution; however, the test was not specifically designed for the evaluation of spray obstructions. Nonetheless, the general direction from the project technical panel (with a few exceptions) was to invest preliminary resources in using ADD testing to explore the role of obstruction severity parameters prior to conducting full-scale tests. Table 1 shows a series of standard test arrangements found in UL Note that only four of these configurations provide enough clearance to install an HVLS fan without enveloping the equipment in the test flame (Tests 1, 2, 5, and 8).

18 Fire Protection Research Foundation 9 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 5. UL 1767 ADD TEST DIAGRAMS [10] TABLE 1. UL 1767 TABLE 30.1 ADD TEST PARAMETER SUMMARY [10] Tests 2 and 8 were identified as being of interest because they offered an opportunity to evaluate single and multiple-sprinkler configurations, respectively at feasible clearances. ESFR sprinklers with a K factor of 14 were used throughout Phase I fire testing for the sake of consistency as well as the availability of historical data for comparison of sprinkler performance. This factor was especially important in the formulation of the experimental approach to full-scale fire testing, which is discussed later in this report.

19 Fire Protection Research Foundation 10 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The intent of investigating both single and multiple-sprinkler configurations was to acknowledge the results of the National Quick Response Sprinkler Research Project, which concluded that the obstruction of a single sprinkler is not a failure point for an ESFR system. Therefore, the single sprinkler test configurations were specified in an effort to gain a fundamental appreciation for the relevant parameters with subsequent evaluation of the multiple-sprinkler case. Additionally, this historical reference cited ignition location between 2 sprinklers as being the most challenging ignition location. Therefore, information regarding the influence of the obstruction located between two sprinklers was desired. The final test plan for ADD Testing is shown below in Table 2. Note that the same fan model is used throughout these tests to produce a range of floating solidity ratios from approximately 0%- 38% simply by moving the same fan to different locations within the fixed spray. Tests 1 and 7 were included to gain baseline results with no fan present. In all other tests, an HVLS fan was proposed to be installed at a position within the clearance space identified in the table. Horizontal offsets between the fan and sprinkler are edge to edge measurements. TABLE 2. PHASE I ADD TEST PLAN Test Total Number of Sprinklers Sprinkler Spacing Horizontal Offset (Fan to Sprinkler) Vertical Offset (Clearance to Sprinkler Deflector) Floating Solidity Ratio 1 1 N/A No Fan No Fan N/A 1 foot 3 feet N/A 4 feet 3 feet N/A 8 feet 3 feet N/A 1 foot 4 feet N/A 1 foot 5 feet Centered Between 2 12 feet No Fan No Fan Centered Between 2 12 feet 6 feet 3 feet 0.19 Full-Scale Fire Testing A number of methods are available to assist with the prediction of various aspects of full-scale fire behavior in a storage or manufacturing facility; though, none are potentially as accurate as the execution of a physical model. However, the obvious drawback to such an approach lies in the economic limitations on the number of configurations that may be studied. For this reason, it is often very useful to include relevant historical data in the analysis where feasible. In this case, the National Quick Response Sprinkler Research Project yielded usable baseline data for K-14 ESFR sprinkler performance under a practical worst-case fire scenario (Test 10). Test 10 was selected as the historical baseline test as it represented the most challenging fire scenario for 20-foot storage under a 30-foot ceiling, which was identified as an important configuration to evaluate in the Phase I research program. This practical worst-case fuel loading consisted of double-row rack storage of Group-A plastics stored to a height of 20-feet underneath a 30-foot ceiling. The length of the array was approximately 32-feet (four 8-foot sections) with 4-foot wide aisles on either side. Heat flux targets in the form of single-row rack storage of the same height and length were provided on

20 Fire Protection Research Foundation 11 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: either side of the aisles. The commodity stored in the targets was Group-A plastic in the middle 2 bays and Class II commodity at either end. Automatic sprinkler protection for this test was provided in the form of K-14 ESFR sprinklers positioned at the ceiling level with 14-inch clearance between the deflector and ceiling and a 10- foot clearance from the top of the storage array to the sprinkler deflector. Sprinklers were spaced across the ceiling in a uniform 10-foot x 10-foot grid with an operating pressure of 50 pounds per square inch (psi). A total of two full-scale fire tests were planned as part of the Phase I research program. Inclusion of the results from the referenced historical fire test as a baseline therefore provided extremely valuable information on which to build toward a final solution. In moving toward this solution from the baseline results, the most logical step for the first full-scale test was to include an HVLS fan positioned within the vertical clearance space between the top of the fuel and the sprinkler deflector. The exact position would be chosen to produce the worst-case spray obstruction as determined by the results of the ADD testing. Additionally, it was determined that the fan should be operating at maximum thrust to produce the worst-case fan-induced airflow conditions on the spread of the fire. In this manner, the first full-scale test was intended to serve as verification (or lack thereof) of the fundamental research problem. Resources for a second full-scale test were reserved to address required fan shut-off time and detection means. Determination of a fan position intended to produce a worst-case scenario for wind-aided fire spread was attempted. This attempt was made with knowledge of the fundamental fluid dynamics of the fan induced airflow in a simple geometry with further insight provided from the results of tests involving HVLS fans and control mode sprinklers [8]. In the simplest geometry with a fan blowing from the top of an empty room, a stagnation flow results as shown in Figure 6 [generated by Aynsley with the use of Phoenics Computational Fluid Dynamics (CFD) software]. The flow of air originating from the fan source is ultimately redirected by the boundaries of the enclosure. In situations where vertical walls are located far from the source and additional fans influence the boundary of the flow, the turbulent mixing that results introduces significant complexity [11]. Similarly, introduction of the storage array significantly disrupts the classic stagnation flow pattern shown in Figure 6 and ultimately acts as a large resistor to fluid flow. FIGURE 6. STAGNATION FLOW PATTERN

21 Fire Protection Research Foundation 12 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: In the referenced test series, it was determined by full-scale test iterations that location of the fire source underneath the tip of the fan blade resulted in faster fire spread than location directly under the fan hub (Figure 7). It should be noted that in those tests, palletized storage of Group A plastic commodity to a height of 15 feet underneath a 25-foot ceiling was tested with a 24-foot diameter fan operating at half speed. FIGURE 7. WORST-CASE FAN LOCATION FOR UL/GE GAP TEST SERIES [8] This worst-case fan placement corresponded to the position producing an airflow that maximized the Required Delivery Density (RDD) by increasing the rate of fire spread. The manner in which this is accomplished within such a complex flow field as that introduced by the rack storage configuration is not well understood. There is some evidence in the literature that the additional heat transfer mechanism introduced by turbulent mixing in the vicinity of the flame and exposed fuel should increase the rate of fire spread, though other parameters may play a similarly important role [12]. The use of computational fluid dynamics to aid in this effort may yield useful information, though a number of obstacles to using Fire Dynamics Simulator (FDS) for such an effort were realized in Phase I. FDS is a Large Eddy Simulation (LES) type model developed by the National Institute of Standards and Technology (NIST). The model is designed to explore thermally driven fluid flow, though the solver could also potentially be used (at an appropriately high resolution) to resolve this forced convective flow field. One of the most significant challenges to using FDS in this application is the resolution of the HVLS fan as a source of fluid flow. A rectilinear approximation of curvilinear fan geometry is necessary with this software, though the consequent inaccuracy introduced into the flow profile of the source is problematic. Additionally, validation of simulated fire spread in rack storage configurations in the absence of HVLS fans is not fully complete. Therefore, simulations would need to focus only on the faninduced airflow in non-fire situations. This is also problematic because the effects of the flow field on fire spread throughout the array are not readily apparent. Instead, parameters such as turbulence intensity become the focus of the results based on limited experimental evidence.

22 Fire Protection Research Foundation 13 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: An attempt was made as part of the Phase I research effort to use FDS in non-fire simulations to assist in determining worst-case fan placement for full-scale Test 1, but adequate resolution of the HVLS fan in particular proved too great an obstacle to overcome with available resources. Fortunately, the experimental lessons learned from the Global Asset Protection Services (GAPS) Research effort did include the observation that ignition location roughly underneath the tip of the fan blades resulted in a worst-case fire scenario for that series. As a result, this fan placement was selected for use in the first full-scale fire test of Phase I. A diagram of the first full-scale test is shown in Figure 8. The rationale for fan placement with respect to sprinkler spacing will be discussed in the analysis of ADD test results. FIGURE 8. DIAGRAM OF FIRST FULL-SCALE FIRE TEST AT UL

23 Fire Protection Research Foundation 14 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: EXPERIMENTAL RESULTS AND ANALYSIS Actual Delivery Density Testing The Phase I ADD test plan was arranged to be executed at Southwest Research Institute (SwRI) in San Antonio, Texas. The official report issued by the test laboratory is contained in Appendix A of this report. The first tests executed in the plan corresponded to the estimated extreme values for floating solidity ratio in the single sprinkler configuration (i.e., 0% and 38%). These were identified as Tests 1 and 2 in Table 2, which is reprinted here for ease of reference. TABLE 2 (RE-PRINTED). PHASE I ADD TEST PLAN Test Total Number of Sprinklers Sprinkler Spacing Horizontal Offset (Fan to Sprinkler) Vertical Offset (Clearance to Sprinkler Deflector) Floating Solidity Ratio 1 1 N/A No Fan No Fan N/A 1 foot 3 feet N/A 4 feet 3 feet N/A 8 feet 3 feet N/A 1 foot 4 feet N/A 1 foot 5 feet Centered Between 2 12 feet No Fan No Fan Centered Between 2 12 feet 6 feet 3 feet 0.19 Although the experimental setup was intended to resemble that which is outlined in UL 1767, certain aspects of the apparatus were unique to the testing laboratory. Perhaps most notably among these features was the method of measuring the rate of water collection. The standard apparatus at UL utilizes a load cell connected to a data acquisition system which measures the weight of water collected at a frequency of 1 Hertz (Hz) throughout each test. The SwRI apparatus did not possess this capability and instead relied on the sequential weighing of each individual collection pan at the conclusion of each test to determine the rate of water collection. For perspective, consider that the sampling rate in this arrangement is approximately 300 Hz, neglecting the time necessary to take the measurements. During this time, the potential for persistent water evaporation from the heated pans should be recognized. In addition to the discrepancy between measuring methods, variability in the degree of insulation of the apparatus ultimately led to the warping of water collection pans and a slight sagging of the supporting steel truss framework as shown in Figures 9 and 10. The slight sagging of the supporting framework apparently occurred as the result of a preliminary calibration test conducted in the laboratory prior to the commencement of the test series. According to laboratory technicians, during this calibration test, the steel framework was left uninsulated from the heptane fire. This carries some significance because of the resulting inward tilt of the water collection pans, which ultimately distorted the size of the flue spaces between them. After this calibration test, the supporting framework was thoroughly insulated to yield a consistent shape for the entire test series. The degree to which the warping of the metal collection pans may have differed from that in the UL apparatus is unclear as the degree of insulation provided for the UL apparatus is not clearly specified in the UL 1767 standard. However, this is important considering that the cross sectional area of the receptacle is affected.

24 Fire Protection Research Foundation 15 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 9. SAGGING OF SUPPORTING STEEL TRUSS FIGURE 10. OBSERVATION OF PAN WARPING

25 Fire Protection Research Foundation 16 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The first test, conducted without an HVLS fan in place, was meant to provide baseline values for rates of water collection from the ESFR spray. The results of this test indicated that a total of 533 gallons of water were collected by the apparatus over a period of 5 minutes. This result does not compare favorably to the UL 1767 criteria, which stipulates a total water collection of 892 gallons for this duration. However, in light of the notable differences in the apparatus highlighted thus far, this significant difference in the measured total water collection is not altogether surprising. Consequently, none of the results obtained in this test series should be compared to ADD test results obtained in any other test series as they are a function of the apparatus used. Ironically, reproducibility was a primary reason for the selection of ADD testing as a means for simple evaluation. The lack of reproducibility ultimately demonstrated in this effort should be noted. Nonetheless, relative comparison of the results within the same series is reasonable due to the consistency of the observed conditions. Having established the baseline values for the apparatus, the next task was to compare extreme value cases, which corresponded to floating solidity ratios of 0.00 (no fan) and 0.38 (fan very close to sprinkler), respectively. The results for water collection rate are compared to those for the baseline case in the standard presentation shown in Table 3. The noted pan locations 1-20 correspond to those denoted in Figure 5 of this report. Pans 4, 7, 10, and 13 are highlighted to draw attention to the significance of their location in the central water collection area. The heavily loaded central core of the ESFR sprinkler spray is often attributed as a critical performance feature. A slight deviation from the standard form of presentation is found in the tabulation of local percent differences between each test. In this representation, negative values are used as a convention to denote a decrease in the measured rate of water collection in the second test with respect to the baseline established in the first. TABLE 3. STANDARD PRESENTATION OF ADD RESULTS (EXTREME CASES) Pan Location No Obstruction - Pan Water Density (gpm/ft 2 ) 38% Obstruction - Pan Water Density (gpm/ft 2 ) % Difference Undef SUM: AVERAGE:

26 Fire Protection Research Foundation 17 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The percentage difference between the values recorded in these extreme cases show essentially no difference between the sum and the average values; however, local changes within the flow field are evident. Comparison of these local variations to an approximation of experimental error was made in an effort to determine their significance. This was done by repeating both tests and comparing identical exercises. The results indicated that a percent change within the range of -12% to 35% could be expected for local pan values with as much as a 13% difference observed in the sum and average quantities. These extraordinarily large error bars are likely due to the manual method of measurement discussed previously in the report with some additional influence from pan warping. Ultimately, several of the water collection rates shown in Table 3 are within this range and therefore can not be clearly distinguished as the result of the flow obstruction or the result of experimental error. Simple observations of shadow effects during the tests indicated that the static HVLS fan caused redistribution of the sprinkler spray (see Figure 11). In this picture, it is important to focus on the lines of water shown at the floor. These lines correspond to the location of the fan blades above and indicate water distribution at the ground level that takes on the same shape as the obstruction, a phenomenon referred to here for ease of reference as flow shadowing. Unfortunately, the form of measuring this phenomenon provided by the test method was not able to capture this phenomenon with anything more than very crude detail. Nonetheless, the objective of the ADD testing was simply to identify a worst-case obstruction configuration and this objective was further pursued by examination of local rates of water collection with more of a focus on local minimums than global sums and averages. FIGURE 11. EXPERIMENTAL OBSERVATION OF FLOW SHADOWING

27 Fire Protection Research Foundation 18 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The relatively poor resolution of the results for comparison of extreme values of floating solidity ratio implied little value for obtaining intermediate data points. Instead, it was hoped that examination of a 2-sprinkler configuration would yield useful information for determining a worstcase static obstruction scenario. Therefore, the test plan proceeded with Tests 7 and 8 as described in Table 2, which focused on the 2-sprinkler configuration with a baseline test (Test 7) and another test with the fan centered horizontally between the sprinklers (Test 8). The results of these tests are tabulated in Table 4. TABLE 4. COMPARISON OF ADD RESULTS FOR 2-SPRINKLER CONFIGURATION Fan at 1 ft. Horizontal, 5 ft. Pan Location No Obstruction - Pan Water Density (gpm/ft 2 ) Vertical Offset - Pan Water Density (gpm/ft 2 ) % Difference SUM: AVERAGE: The results again illustrate a pronounced disruption of the ESFR spray throughout the collection area. Figure 12 was generated in an effort to evaluate the severity of local disturbances in this 2-sprinkler case versus the 1-sprinkler case. The data in this plot compares the absolute value of the percentage difference between the obstructed and unobstructed tests for both the single and double sprinkler configurations. Although it is clear that significant disruptions are measured throughout the flow field, this comparison does not definitively assert the severity of one configuration over another. Half of the collection pans appear to be more severely influenced in the 1-sprinkler configuration and vice versa for the other half. Given the particularly large error bars discovered early in the test program, further comparison of the results on this basis appears to offer little value.

28 Fire Protection Research Foundation 19 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 12. EVALUATION OF FLOW REDISTRIBUTION Absolute Value of % Difference (Obstructed versus Unobstructed) Sprinkler Configuration 2-Sprinkler Configuration Numbered Collection Pans However, it is worth noting that in this 2-sprinkler configuration, when the fan is in place, 25% less water accumulates in the collection area over the duration of the test. This result did not occur in the case of the 1-sprinkler configuration, where nearly identical total amounts of water were collected in the obstructed versus unobstructed tests. This suggests that in the 2-sprinkler configuration, the obstruction actually redirected a substantial amount of water beyond the boundaries of the collection pans. The likely reason for this is that many droplets with initial trajectories toward the collection area were redirected by accumulating on and subsequently cascading off of the obstruction. Ultimately, this is evidence of interference in the overlapping of adjacent water distribution patterns. NFPA 13 currently does not allow for such interference to occur closer than 3 feet below the sprinkler deflector. In this case, the vertical distance criterion is satisfied, but the extent of the challenge to the full-scale protection scheme is unclear. The final issue of interest for investigation in the ADD test series was the effect of changing the vertical clearance between the fan and the sprinkler spray. The major HVLS models under investigation in this Phase I test series all possess manufacturer recommended clearances between 3-feet and 5-feet below the finished ceiling. For this reason, the final ADD test was conducted to investigate the effect of changing the vertical clearance from 5-feet to 3-feet below the finished ceiling. The results for this test are shown in Table 5. The data appears to indicate a change in the spray distribution that is generally less pronounced than the other scenarios tested.

29 Fire Protection Research Foundation 20 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: TABLE 5. COMPARISON OF ADD RESULTS FOR CHANGE IN VERTICAL CLEARANCE Fan at 1 ft. Horizontal, 3 ft. Vertical Offset - Pan Water Fan at 1 ft. Horizontal, 5 ft. Vertical Offset - Pan Water Pan Location Density (gpm/ft 2 ) Density (gpm/ft 2 ) % Difference Undef SUM: AVERAGE: Ultimately, very little useful information was obtained in evaluating obstruction severity using what was ultimately found to be a non-standard apparatus to conduct standard ADD testing. The initial hypothesis that obstruction severity increases with increasing size and closer proximity of the obstruction to the sprinkler was neither proven nor disproven by the approach. No information was obtainable on droplet size or velocity using this test method. These are both very important in the determination of the efficiency of a particular flow pattern. Perhaps the only significant piece of information obtained was the measurement of a 25% change in total water collection apparently caused by the fan obstruction in the 2-sprinkler configuration. Revisiting the conclusions of the National Quick Response Sprinkler Research Project, it is important to recall that full obstruction of a single sprinkler was determined not to be a failure point of an ESFR system. The key to system failure was the disruption of multiple sprinklers. The ADD testing conducted in the current program revealed the potential for HVLS fans to significantly redistribute a sprinkler spray in several different configurations. Yet, deciding which configuration most threatens system success should take this historical data into account. It was ultimately decided that the HVLS fan should be placed at a minimum vertical clearance of 3-feet below sprinkler deflectors and a horizontal placement between 4 sprinklers in full-scale testing. Placement between 4 sprinklers was not investigated in the ADD testing due in part to lack of a workable clearance in the standard 4-sprinkler configurations. However, the results of the 2-sprinkler tests, where the fan is placed 6-feet away from each sprinkler, demonstrated significant redistribution of both flow fields. In a 10-foot by 10-foot sprinkler grid, this 6-foot distance is very close to the distance resulting from placement between either 2 or 4 sprinklers.

30 Fire Protection Research Foundation 21 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: Therefore, in an effort to disrupt the entire first ring, placement between 4 was selected. Additionally, this fan placement was selected to simplify the obstruction problem and was considered by the fan manufacturers as a commonly used and practical fan placement. Full-Scale Fire Testing The Phase I full-scale fire test plan was arranged to be executed at Underwriters Laboratories (UL) in Northbrook, Illinois. The official report issued by the test laboratory is contained in Appendix B of this report. The objective of the first full-scale fire test was simply to verify the problem with HVLS fans operating over high hazard storage configurations (20-foot storage height under a 30-foot ceiling) that are protected by ESFR sprinklers at the ceiling level only. In order to determine whether the objective was met, both the first and second tests were designed for comparison to Test 10 of the National Quick Response Sprinkler Research Project as a control. The pass/fail criteria used for the full-scale testing, as defined in the National Quick Response Sprinkler Research Project, is that early suppression is achieved when: Not more than four sprinklers activate when the sprinklers open in the proper sequence. Ceiling gas temperatures are such that exposed structural steel would not be endangered (less than 1,000 o F throughout the entire test). Fuel consumption, for double row racks with four tiers (20-feet storage height), does not exceed 4-4 ¾ equivalent pallet loads. It is important to note that the acceptable number of operating sprinklers, four, applies only when sprinklers operate in proper sequence all four of the first ring sprinklers operating within a few seconds and definitely before any second or third ring sprinklers. Should a different operating sequence occur, as many as 8-10 sprinklers may operate and early suppression still be achieved if the other criteria is met [6]. Additionally, when ESFR sprinklers are tested under more challenging situations, early suppression is not always expected [6]. The results of such a test may resemble fire control in lieu of suppression, in which ceiling temperatures were within the acceptable range, however, the number of operating sprinklers exceeded four and the amount of damage exceeded 4-4 ¾ pallets for the four-tier array. An example of such an outcome can be found in the National Quick Response Sprinkler Research Project Test 6. In this test one of the first ring sprinklers was plugged causing 11 sprinklers to operate and the fire to consume six pallets, one of which was across the aisle from the ignition array. Ceiling temperatures were acceptable however. Plotting the ceiling temperatures above ignition versus time reveals a curve which more reassembles a fire control scenario which can be typified as a period of fire regrowth after the first ring of sprinklers activate followed by activation of second ring sprinklers and a subsequent reduction in the heat release rate of the fire. During the control test, fire suppression was achieved with the activation of only 3 sprinklers and a total equivalent load of 3.5 pallets was consumed without aisle jump. Observers of the test noted that flames reached the top of the array (presumably via the flue space) within 35

31 Fire Protection Research Foundation 22 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: seconds after ignition with the first sprinkler operating at 61 seconds past the ignition time. A secondary path of fire spread was within the shielded clear space separating the first and second tiers of storage. The only significant variable change introduced in the new test was the presence of an HVLS fan (24-foot diameter Whisperfoil XL manufactured by MacroAir) operating at a maximum speed (63 rotations per minute) throughout the entire duration of the test. Almost identical observations of early fire growth were documented including the primary and secondary paths of fire spread up the flue and within the clear space between the lowest storage tiers (Figure 14). Very shortly after flames were observed protruding through the top of the array and bending away from the fan, activation of the two sprinklers closest to ignition occurred. These activations were at 60 and 64 seconds after ignition, respectively, as shown in Figure 15. FIGURE 14. OBSERVATIONS OF FIRE SPREAD IN FULL-SCALE TEST 1 Rotating HVLS Fan Shielded fire spread between storage tiers

32 Fire Protection Research Foundation 23 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 15. UL DIAGRAM OF SPRINKLER ACTIVATIONS FULL-SCALE TEST 1 Ultimately, the water from these sprinklers forced the fire to retreat to burning within the somewhat shielded clearance space between storage tiers. However, at this early stage in fire development, the reaction had not gained sufficient inertia to overcome the rate of water delivery to even these partially shielded areas. As a result, suppression was achieved without the activation of any additional sprinklers or the occurrence of aisle jump. The peak gas temperature recorded at the ceiling above the location of ignition was 250 F with a peak steel temperature of 109 F. These results demonstrate fire suppression by the activation of only 2 sprinklers and consequently compare quite favorably to the established pass/fail criteria. Additionally, the fundamental fire protection problem presented by the operating HVLS fan on the ESFR sprinkler system performance is called into question by comparison to the historical baseline test results. Recall that the baseline test required one additional sprinkler to yield fire

33 Fire Protection Research Foundation 24 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: suppression. The total number of equivalent pallet loads consumed was approximately 3.5 in both the baseline test and Test 1 of the current series. The results of the first full-scale fire test indicated that what was thought to be the worst-case fire scenario did not significantly challenge the performance of the ESFR sprinkler system. Consequently, the rationale behind selection of this fire scenario was revisited. In this selection, emphasis was placed on the relative locations of the fan, ignition, and ceiling level sprinklers. Observations during the test revealed that the rack-storage array significantly disrupted the faninduced airflow, thereby minimizing its influence on early fire spread. This observation was markedly different from observations during the UL/GE GAPS testing effort in which palletized storage was used in a different configuration (i.e., storage height, clearance, etc.) [8]. This suggests that worst-case fan placement relative to ignition is a strong function of storage geometry and potentially other boundary conditions (i.e., walls, additional HVLS fans, etc.). In other words, placement of ignition underneath the edge of the fan does not guarantee a worstcase fire scenario. Revisiting the fundamentals of ESFR sprinkler performance, the most challenging fire scenario is one in which the RDD is maximized. Observations of early fire growth within the rack storage array during Test 1 indicated that this could potentially be achieved by encouraging fire spread within the shielded space between storage tiers while simultaneously discouraging fire growth toward the top of the array. The goal then of the second full-scale fire test was to design a fire shielded from ESFR sprinklers by the fan-induced airflow (Figure 16). This was approached by positioning the fan (still rotating at 63 revolutions per minute (rpm) throughout the test) centered directly above the point of ignition. In an effort to maintain the severity of the spray obstruction, this location corresponded to a spot between 4 sprinklers as shown in Figure 17. All other experimental variables remained the same as in Test 1. FIGURE 16. EARLY FIRE GROWTH WITHIN LOWER TIERS TEST 2

34 Fire Protection Research Foundation 25 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 17. UL DIAGRAM OF FULL-SCALE TEST 2 SETUP Ultimately, the second full-scale test did prove to be a more challenging fire scenario for the ESFR system for apparently the reasons identified in the above discussion. This fire gained intensity within the lower tiers of the array where it was shielded from the sprinkler discharge by the storage arrangement. Evidence of this early fire growth within the lower tiers is shown in Figure 16. A total of eight ESFR sprinklers activated within the first five minutes after ignition (see Figure 17), though the total duration of the test extended an additional 27 minutes. At the time of test termination, residual burning was actively suppressed. Review of the sprinkler operating sequence shows that the first three sprinklers to activate were in the first ring, followed by four sprinklers in the second ring. The activation of the referenced first-ring sprinklers occurred

35 Fire Protection Research Foundation 26 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: within approximately the same time frame as the first full-scale test; however, the performance of these sprinklers was clearly hindered by the presence of the operating fan as their response was not sufficient to yield suppression. A comparison of the results to the pass/fail criteria shows that ceiling temperatures were within acceptable levels (peak gas and steel temperatures at the ceiling were 237 F and 145 F, respectively). The fire did consume approximately 10 pallet loads, however the damage was generally confined to the two bays above the ignition location. Based on these observations and an analysis of damage diagrams, it appears that the fire was controlled by the ESFR system during the testing timeframe. CONCLUSION AND RECOMMENDATIONS The objective of the Phase I research effort was to explore the interaction between a High Volume Low Speed (HVLS) fan and Early Suppression Fast Response (ESFR) sprinkler protection of rack and palletized commodity storage. This interaction was specifically defined in terms of the obstruction of sprinkler discharge and the required fan shut-off time, including detection means. The obstruction of sprinkler discharge was investigated in a series of Actual Delivery Density (ADD) tests designed in accordance with UL These tests were performed to evaluate the hypothesis that obstruction severity would increase with increasing size and proximity of the static fan obstruction to the sprinkler. The hypothesis was neither proven nor disproven by the test method, which provided very little information regarding the role of the obstruction in manipulating droplet trajectories. Additionally, droplet sizes and velocities are known to be important characteristics of the distribution pattern. No quantifiable information regarding these parameters can be provided by standard ADD testing. Literature review revealed that full obstruction of a single ESFR sprinkler (plugged sprinkler) is not a failure point for the system. With this in mind, the search for a worst-case fan placement relative to sprinklers focused on the obstruction of multiple devices. The results of the ADD tests indicated significant interference of the HVLS fan with sprinkler overlapping in a 2-sprinkler configuration. It is known that such overlapping, particularly in the first ring of sprinklers, is important to the success of the ESFR system. Therefore, fan placement between four sprinklers was pursued in subsequent full-scale fire testing. Additionally, this fan placement was selected to simplify the obstruction problem and was considered by the fan manufacturers as a commonly used and practical fan placement. A total of two full-scale tests were conducted at Underwriters Laboratories following the completion of the ADD effort. Both tests were performed using ESFR protection of rack storage of Group-A plastics stored to a height of 20-feet underneath a 30-foot ceiling. The first major objective was to verify the existence of a problem with an HVLS fan operating at maximum thrust and positioned to achieve the worst-case obstruction of sprinkler discharge. Fan placement was varied between tests with the worst-case results measured for fan placement centered over the point of ignition. The results apply only to this specific storage configuration as different boundary conditions will likely result in a different influence of the airflow on fire spread. It is thought that these results were more severe than the case of ignition below the tip

36 Fire Protection Research Foundation 27 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: of the blade because the former strategy promotes shielded fire growth in the space between storage tiers. Evaluation of the full-scale test results may be made in terms of the established pass/fail criteria as well as comparison to a historical baseline test conducted as part of the National Quick Response Sprinkler Research Project. Results of the two full-scale fire tests revealed no threat to the structure with ceiling gas temperatures briefly peaking at values on the order of 250 o F for both tests. A total of 8 sprinklers activated in the test in which the fan was positioned directly above the point of ignition (Test 2). This is significantly higher than either the first full-scale test (Test 1) or the baseline case, in which fire suppression was achieved with the activation of 2 and 3 sprinklers, respectively. The increased number of sprinklers which activated in Test 2 may be attributed to the sequence of sprinkler operation as second-ring sprinklers operated prior to the four first-ring sprinklers. The National Quick Response Sprinkler Research Project report suggests activation of 8-10 sprinklers is not uncommon in that circumstance. Nevertheless, the total number of sprinklers activated during the second full-scale test was definitively less than 12, which is an established design criterion for ESFR systems. Activation of the first sprinkler was consistently at or very near one minute after ignition in all three cases. Transient ceiling temperatures located directly over ignition often mimic the rate of fire growth and therefore may be used to assess performance of the ESFR system. The convective currents introduced by the presence of an operating HVLS fan complicates any efforts to draw distinct correlations between these parameters; however, evaluation of these temperatures over extended time-scales still provides useful indicators of general fire growth or decay. In Test 1 these temperatures are suppressed after the activation of the first two sprinklers, implying suppression at the seat of the fire directly below. The response of these same gas temperatures in Test 2 was noticeably different. In this case, a shallow decline after the nearly simultaneous activation of the first three sprinklers is followed by slight re-growth. A gradual decay ultimately results as additional sprinklers activate. Water from sprinklers generally serves to absorb heat from gaseous flames and solid surfaces as well as limiting further fire spread by wetting unburned commodity surrounding the ignition site. The more gradual decline in ceiling gas temperatures in Test 2 suggests that this process occurred in a less efficient manner than Test 1. The timing of the sharp decline in ceiling temperatures in Test 1 is consistent with the definition of fire suppression given in NFPA 13. In contrast, the results of Test 2 appear to be more consistent with the definition of fire control given by the standard. As previously discussed, such performance from ESFR sprinklers is considered acceptable for the challenges presented in Tests 1 and 2. The authors then conclude that the presence of the HVLS fans did not affect the performance of the ESFR sprinklers to a level that would be considered unacceptable and that the performance of the sprinkler system is consistent with the criteria established in the baseline testing project. As a result of the Phase I research experiments, measurements and observations discussed in this report, the following conclusions and recommendations are made regarding obstruction of sprinkler discharge and the issue of fan shutoff: 1. Phase I results apply only to 30-foot high facilities with 20-foot high storage, which encompasses a significant portion of current HVLS storage and manufacturing applications.

37 Fire Protection Research Foundation 28 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: The minimum vertical clearance between the fan obstruction and a sprinkler deflector at ceiling level should be 3 feet as currently allowed by NFPA 13 for clearance to storage. The effects of reduced vertical clearances of the obstruction (below 3 feet) were not explored in this research effort. 3. HVLS fans were consistently placed between 4 sprinklers in all full-scale tests conducted in Phase I. In an effort to build on the results presented in this report, it is recommended that HVLS fans be installed between 4 sprinklers. ADD testing was unable to definitively determine a worst-case fan placement relative to nearby sprinklers. Particle Image Velocimetry (PIV) and Phase Doppler Interferometry (PDI) techniques potentially offer a means for obtaining definitive obstruction severity data, though these approaches are currently not standardized and also require a significant investment of resources. 4. The influence of fan-induced airflow on fire spread is a strong function of boundary conditions such as fan airflow (376,804 cfm used in Phase I) storage geometry, clearance from fan to storage, height of the test facility, distance to walls and the location of additional fans. In the configuration studied in Phase I, there is evidence that the storage arrangement significantly buffers the fan-induced airflow and minimizes its influence on fire spread. Based on this evidence, it was assumed that the worst-case placement of the fan relative to ignition in this configuration was directly above the latter. Further research is recommended to verify that this is indeed the worst-case placement of the fan relative to ignition. Such research could be conducted by a number of means including computational modeling [obstacles to using Fire Dynamics Simulator (FDS) noted in this report], scale modeling or additional full-scale testing. 5. The fan model chosen for testing in the Phase I research effort possesses a central solidity ratio of 0.60 as illustrated in Figure 3 of this report. According to this characterization, 83% of the fan models analyzed in this report possess nearly identical or less invasive shapes than the model tested in Phase I. 6. Successful system performance was achieved without fan shutoff in both full-scale tests. An attempt was made in Phase I to evaluate ESFR system performance in full-scale fire testing under practical worst-case fire conditions. The basis for generating these conditions relied heavily on the positioning of the fan relative to sprinklers and ignition location. It is clear that the flow field is strongly influenced by boundary conditions such as the geometry of the storage array, the clearance between the fan and storage, clearance from fan to room boundaries (i.e., walls and floors) and spacing between multiple operating fans where applicable. It is therefore recommended that a thorough analysis of the relevant physics be performed to determine characteristic fluid flow parameters responsible for enhancing fire spread (i.e., turbulence intensity). Such parameters could then be explored in non-fire airflow testing and confirmed in full-scale fire testing. The objective of this effort would be to either confirm or challenge fan placement relative to the location of ignition in the Phase I tests. To that end, a suggested test plan for Phase II of this program is provided within this report.

38 Fire Protection Research Foundation 29 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: RECOMMENDATIONS FOR PHASE II RESEARCH A major objective of the full-scale fire testing conducted during Phase I of the research program was to determine the maximum fire challenge to ceiling-level ESFR sprinkler protection when HVLS fans operate over 20-foot tall rack-storage in a 30-foot tall warehouse. Some assumptions were made in formulating the full-scale scenarios thought to represent this maximum challenge. The most significant among these was the assumption that placement of the fan with respect to the ignition location optimized wind-aided fire spread. Validation of this assumption is important for strengthening the conclusion of Phase I. Additionally, a roadmap for Phase II may be developed if a deeper understanding of the parameters governing optimization of wind-aided fire spread is achieved. Such knowledge will provide critical insight in the selection of pertinent configurations for full-scale fire testing. Once a set of appropriate configurations is determined for Phase II, focus may be placed on the response of the automatic sprinkler system and automatic detection means. Prior to Phase I, full-scale fire testing conducted by UL/GE GAPS (Test 1) revealed at least one storage application where wind-aided fire spread caused by the HVLS fan was apparently responsible for the failure of control-mode sprinkler protection. However, full-scale fire testing conducted in Phase I revealed less severe effects for a different storage configuration protected by a more aggressive (ESFR) system. By first seeking detailed insight into the parameters dictating optimal wind-aided fire spread, Phase II seeks to answer the question of whether this contrast in results is due primarily to airflow or to the performance of the sprinkler system. Therefore, it is proposed that a series of experiments designed to examine airflow parameters be followed by full-scale fire tests focusing on the performance of sprinklers under optimal wind-aided fire spread conditions. Throughout this testing, it is proposed that information be gathered on the parallel performance of a number of common detection technologies including heat, smoke, and carbon monoxide detection. Comparative hazard-specific data could then be used as a framework for evaluating practical and effective means for initiating action by a fire alarm system, should it be deemed a necessary component to the overall fire protection approach. The contrast in the full-scale fire test results between Phase I and the UL/GE GAPS Test 1 emphasizes the need for a more precise understanding of the role of boundary conditions, such as facility height, fan clearance to walls or adjacent fans, fan clearance to storage, and the locations and sizes of aisles and flues, in dictating the spread of fire and combustion products. A number of modeling approaches potentially represent a means for acquiring such data in relatively inexpensive fashion; however, such approaches would at some point require validation by full-scale measurements. For this reason, it is recommended that Phase II experimentation begin with full-scale measurements of air velocities throughout the flow domain as a function of varying boundary conditions. Such measurements would be taken in the absence of fire and without the activation of automatic sprinklers. The benefits of this approach include the following: 1. Full-scale experimental measurements do not require validation by another method. 2. Cost is limited by removing the destructive parameters of fire and water. 3. Full-scale measurements provide a means for validation of future modeling efforts.

39 Fire Protection Research Foundation 30 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: Information is obtained on airflow within the array as well as near the ceiling, the latter of which will be of great significance in the evaluation of detection/shutdown mechanisms. A movable ceiling in the laboratory provides versatility that allows for full-scale exploration of the role of vertical clearances (fan-to-floor and fan-to-storage) with relative ease. For this reason, it is recommended that a facility with this capability be selected for these tests. Altogether, a total of nine full-scale configurations are proposed for evaluation as outlined in Table 6. The configurations outlined in this table include full-scale Test 1 conducted by UL/GE GAPS as well as both full-scale tests conducted in the Phase I research effort. The intent of this comparison is to exploit the role of fan-induced airflow as a factor in producing the contrasting results of these efforts. Additional measurements are taken in the rack-storage configuration for varying vertical clearances and horizontal offsets between the fan centerline and the ignition location. The measurements taken in the proposed testing configurations shown in Table 1 will provide insight on the relative significance of various boundary conditions on flow velocities near the ignition location and at the ceiling. This insight may be used to confidently identify worst-case fan placement with respect to ignition in a variety of configurations. Consequently, more expensive full-scale fire testing may be reserved for exploration of other parameters. TABLE 6. PROPOSED AIRFLOW EXPERIMENTATION Test Storage Type Ceiling Height Storage Height Horizontal Offset (Fan to Ignition) Notes 1 Solid-piled, palletized 25 feet 15 feet Fan radius Same as UL/GE GAPS Test 1 2 Rack 30 feet 20 feet No offset Same as Phase I Test 1 3 Rack 30 feet 20 feet No offset Same as Phase I Test 2 4 Rack 30 feet 15 feet No offset Increase clearance to storage 5 Rack 40 feet 30 feet No offset Increase ceiling height 6 Rack 30 feet 20 feet No offset Reduce room area with walls 7 Rack 30 feet 20 feet Fan radius Fan centerline in plane with ignition longitudinal flue centerline 8 Rack 30 feet 20 feet Fan radius Fan centerline in plane with ignition transverse flue centerline 9 Rack 30 feet 20 feet Fan radius Fan centerline in plane with aisle centerline It is anticipated that significant insight will be provided by the results of the airflow experimentation outlined in Table 6. With this information in hand, the logic for proceeding with further full-scale fire tests is provided in Figure 18. The logic tree shown in Figure 18 begins with the identification of worst-case fan position relative to ignition location. Subsequent testing is characterized by an or gate, which is depicted to show that the results of airflow testing may either confirm or challenge fan placement relative to ignition in Phase I. If Phase I fan placement is challenged, the newly discovered worst-case scenario will be tested. Otherwise, Test 2 of Phase I will be repeated. The purpose of this repetition is to investigate whether sprinkler operation out-of-sequence in the first run of this test is repeatable and therefore fan-induced.

40 Fire Protection Research Foundation 31 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: Two additional full-scale fire tests are then recommended based on the following rationale: 1. Investigation of taller storage/manufacturing facilities. The most common configuration is 30 feet of rack-storage within a 40-foot tall enclosure. In order to compare the results of this test to Phase I results, this test will be designed to mimic Test 2 from Phase I with the exception of the change in storage and facility height. Using K-14 sprinklers at this elevation would also require an increase in operating pressure to 75 psi. This does represent a change in multiple variables from the referenced test. Therefore, one approach is to allocate funds for multiple full-scale fire tests in exploration of this topic. As an alternative, a single full-scale fire test may be performed with supplemental information provided by the results of the airflow experimentation. This supplemental information may be used to assess the significance of the change in storage clearance. 2. An increase in the clearance to storage should be explored due to the potential decrease in heat collection near the ceiling in this configuration. In order to facilitate simple comparison to established test results, it is recommended that the test configuration include a 30-foot tall space with storage to a height of 15 feet.

41 Fire Protection Research Foundation 32 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: FIGURE 18. LOGIC TREE FOR PHASE II FULL-SCALE FIRE TEST SERIES Use results of airflow testing to identify worst-case fan position relative to ignition location or Repeat Phase I, Test 2 (ESFR Sprinklers) Test newly-discovered worst-case fan position in Phase I test configuration (ESFR Sprinklers) Conduct new full-scale fire test with worst-case fan placement relative to ignition in a 40-foot tall facility with 30-foot storage height. (ESFR Sprinklers) Explore effects of increasing clearance to storage. Conduct new full-scale fire test with worst-case fan placement relative to ignition in a 30-foot tall facility with 15-foot storage height. (ESFR Sprinklers) Moving forward it is also recommended that funds be reserved for an additional three full-scale fire tests using control mode sprinklers for comparison to ESFR performance. The starting point of this testing will be a repeat of Test 2 from Phase I, using control mode sprinklers. An estimation of the costs associated with the scope outlined in these recommendations is provided in Appendix D of this report.

42 Fire Protection Research Foundation 33 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: Submitted by, SCHIRMER ENGINEERING CORPORATION SCHIRMER ENGINEERING CORPORATION Jonathan Perricone, P.E. Associate Engineer Garner A. Palenske, P.E. Vice President/Regional Engineering Manager JP:jm:kh S:\SNDGO\PROJECTS\JOB 2008\ FPRF HVLS Fans and Sprinkler Interaction\rpjp.124.HVLS Final Report doc

43 Fire Protection Research Foundation 34 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: BIBLIOGRAPHY 1. Photos courtesy of MacroAir Technologies, LLC 2. Isman, K. Water Based Fire-Suppression Equipment, Fire Protection Handbook, 20 th edition, Section 16, Chapter 1, p National Fire Protection Association: MA, FM Global Property Loss Prevention Data Sheets 2-2: Installation Rules For Suppression Mode Automatic Sprinklers, September Dubay, C. Automatic Sprinkler Systems Handbook, ninth edition. National Fire Protection Association, MA: Carey, William, Early Suppression Fast Response Sprinklers A New Technology, LabData, Volume 16, No.4, 1985, pp Chicarello, Troup and Dean. The National Quick Response Sprinkler Research Project: Large Scale Fire Test Evaluation of ESFR Automatic Sprinklers, Fire Protection Research Foundation Report, May Yao, Cheng. The Development of the ESFR Sprinkler System. Fire Safety Journal, 14 (1988) Willse, P. and Pabich, M. Fire Test Evaluation on the Effects of High Volume Low Speed Fans on Ceiling Sprinklers, Global Asset Protection Services Research: Internal Publication and Presentation at 2008 NFPA World Safety Conference & Exposition. 9. Widmann, J., Sheppard, D. and Lueptow, R. Non-Intrusive Measurement in Fire Sprinkler Sprays, Fire Technology, Vol. 37, No. 4, 2001, pp Beyler, C. The Effects of Selected Variables on the Distribution of Water from Automatic Sprinklers, Volume 1: Analysis, Department of Fire Protection Engineering, University of Maryland, College Park, MD, Valentine, V. and Isman, K. Interaction of Residential Sprinklers, Ceiling Fans and Similar Obstructions. Fire Protection Research Foundation Report, January Luna, M. and Trevino, J. Report of Special Project Testing of Big Ass Fan s Powerfoil TM 20-ft Ultra Fan Assembly to its affect on the Actual Delivered Density of a K = 14 Sprinkler Head to a modified UL 1767 Test, Intertek Report Number , May 29, Viking Corporation Quintiere, J. Fundamentals of Fire Phenomena. John Wiley & Sons, Ltd. Chichester, UK: p Bryan, J. Automatic Sprinkler and Standpipe Systems, 3 rd edition. National Fire Protection Association: MA, p. 300.

44 Fire Protection Research Foundation 35 February 17, 2009 HVLS Fans and Sprinkler Operation Phase I SEC Project No.: UL 1767 Standard for Early-Suppression Fast-Response Sprinklers, 2005 edition. 17. Aynsley, R. and Ali, M. Optimizing Ceiling Fan Locations with CFD, Architectural Engineering 2003 Conference Proceedings, USA, September Zhou, L. Solid Fuel Flame Spread and Mass Burning in Turbulent Flow, NIST GCR ; 232 p. March 1992.

45 APPENDIX A SOUTHWEST RESEARCH INSTITUTE FIRE TEST REPORT ADD TESTING

46 SOUTHWEST RESEARCH 8220 CULEBRA RD P.O. DRAWER SAN ANTONIO, TEXAS, USA (210) tt CHEMISTRY AND CHEMICAL ENGINEERING DIVISION FIRE TECHNOLOGY DEPARTMENT FAX (210) September 29,2008 Ms. Kathleen Almand The Fire Protection Research Foundation 1 Batterymarch Park Quincy, MA Subject: Southwest Research Institute (swfu@) Project No O1.OO 1 Consisting of 19 Pages Dear Ms. Almand: TEST DATA This letter summarizes work conducted by Southwest Research Institute (SwRI) for The Fire Protection Research Foundation, located in Quincy, MA, on September 5,2008. Eight tests were conducted in general accordance with Section 30 of UL 1767, Early-Suppression Fast-Response Sprinklers, Actual Delivered Density (ADD) Test. The tests were conducted in general accordance, since most of the tests were conducted with a high-volume low-speed fan installed over the apparatus. In addition, only two of the test configurations (Test #2 and 8 from Table 30.1), as outlined in Section 30 of UL 1767, were considered in this evaluation. Appendix A contains a summary of the results from the UL 1767 (ADD) tests, and Appendix B contains heptane mass flow and water pressure data from all the tests. The results apply specifically to the specimen tested, in the manner tested, and not to similar materials, nor to the performance when used in combination with other materials. This document fulfills all contractual deliverables associated with this project. If you have any questions or if I can be of further assistance, please feel free to contact me by phone at , by fax at , or by at jason.huczek@swri.org. Sincerely, Approved, a Senior Research Engineer Engineering and Research Section Director Fire Technology Department This repori Is for the infomalion d M dirt. This repon shall not be reproduced ex=@ in full,!#itha* the ~tittn approval of SwRI. 1 Neither this repon nor the name of the Institute shall be used in publidty or ndvertlsing. I HOUSTON, TEXAS (713) WASHINGTON, DC (301)

47 APPENDIX A SUMMARY OF UL 1767 (ADD) TEST RESULTS (CONSISTING OF 9 PAGES) The Fire Protection Research Foundation SwRI Project No

48 Table A-1. Summary of ADD Test Data (Commodity Pans). A Average 16-Pan Sum of Average Individual 1 Test No. Water Density Pan Densities * No Fire (Water Collection Only) Table A-2. Summary of ADD Test Data (Flue Pans). * No Fire (Water Collection Only) The Fire Protection Research Foundation SwRI Project No

49 Table A-3. ADD Test 1: Baseline (I Sprinkler). Flue-South Flue-East I Sum Average , The Fire Protection Research Foundation SwRl Project No

50 Table A-4. ADD Test la: Baseline (1 Sprinkler - No Fire). Flue-East I Sum Average The Fire Protection Research Foundation SwRl Project No

51 Table A-5. ADD Test 2: 1 Sprinkler (Horizontal Fan Offset: lft; Vertical Fan Offset: 3 ft). The Fire Protection Research Foundation SwRl Project No

52 Table A-6. ADD Test 3: REPEAT: 1 Sprinkler (Horizontal Fan Offset: 1 ft; Vertical Fan Offset: 3 ft). The Fire Protection Research Foundation SwRI Project No

53 Table A-7. ADD Test 4: REPEAT: Baseline (1 Sprinkler). The Fire Protection Research Foundation SwRI Project No

54 Table A-8. ADD Test 5: Baseline (2 Sprinklers). Flue-South Flue-East Sum Average The Fire Protection Research Foundation SwRI Project No

55 Table A-9. ADD Test 6: 2 Sprinklers (Horizontal Pan Offset: 6 ft; Vertical Fan Offset: 3 ft). The Fire Protection Research Foundation SwRl Project No

56 Table A-8. ADD Test 7: 2 Sprinklers (Horizontal Fan Offset: 6 ft; Vertical Fan Offset: 5 ft). The Fire Protection Research Foundation SwRI Project No

57 APPENDIX B SUMMARY OF UL 1767 (ADD) TEST RESULTS (CONSISTING OF 7 PAGES) The Fire Protection Research Foundation SwRl Project No OO 1

58 The Fire Protection Research Foundation SwRI Project No Test Date: September 3,2008 Test ID: fprf-1 40 Test 1: Baseline (1 Sprinkler) Heptane Flow -Water Pressure - s 20 k a 2-8 C, Average Heptane Flow of 9.17 Iblmin = 122, BTUImin, assuming a 0.3 radiative fraction for heptane I I I, A, Time (min)

59 The Fie Protection Research Foundation SwRI Project No Test Date: September 3,2008 Test ID: fprf-2 40 Test 2: 1 Sprinkler (Horizontal 0ffset:l ft; Vertical Offset: 3 ft) Heptane Flow -Water Pressure - h Average Heptane Flow of 8.24 Iblmin = I 10,351 BTUImin, assuming a 0.3 radiative fraction for heptane I Time (min)

60 The Fire Protection Research Foundation SwRI Project No Test Date: September 4,2008 Test ID: fjprf-3 40 L 3 5 Test 3: REPEAT: 1 Sprinkler (Horizontal 0ffset:l ft; Vertical Offset: 3 ft) Heptane Flow 1 -Water Pressure - - Average Heptane Flow of 8.26 Iblmin = 1 10,618 BTUImin, assuming a 0.3 radiative fraction for heptane. Time (min)

61 The Fire Protection Research Foundation SwRI Project No Test Date: September 4, Heptane Flow -Water Pressure - 70 a2 9 C, Average Heptane Flow of 8.22 Iblmin = 11 0,083 BTUImin, assuming a 0.3 radiative fraction for heptane. Time (min)

62 The Fire Protection Research Foundation SwRI Project No Test Date: September 4,2008 Test ID: fprf-5a Test 5: Baseline (2 Sprinklers) Heptane Flow 35 -Water Pressure : n 25 \ 2 w & S 20 k a 9.c, 15 V I Average Heptane Flow of 8.19 Iblmin = 109,681 - BTUImin, assuming a 0.3 radiative fraction for heptane B CI z 0 40 g c 'l (r, h V Time (min)

63 The Fire Protection Research Foundation SwRI Project No Test Date: September 5,2008 Test ID: fprf-6 Test 6: 2 Sprinklers (Horizontal Offset: 6 ft; Vertical Offset: 3 ft). 40 L J Heptane Flow -Water Pressure - 70 r n E: \ s w s ki Q) c Q.c, Average Heptane Flow of 8.35 Iblmin = 111,824.,$' 15 BTUImin, assuming a 0.3 radiative fraction for heptane ' 1 ~ ( ' ~ ~ ~ I 4 Time (min)

64 The Fire Protection Research Foundation SwRI Project No Test Date: September 5,2008 Test ID: fprf-7 Test 7: 2 Sprinklers (Horizontal Offset: 6 ft; Vertical Offset: 5 ft). L - Heptane Flow -Water Pressure J ~ I,,,, I I I I J I I I I # I I I O I,,, Time (min)

65 APPENDIX B UNDERWRITERS LABORATORIES FIRE TEST REPORT FULL-SCALE TESTING

66 Nominal K=14.0 ESFR Ceiling Sprinklers Protecting 20-Ft. High Rack Storage of Group A Plastic Commodity Under a 30-Ft. Ceiling with an Operating High Volume Low Speed (HVLS) Fan Technical Report Underwriters Laboratories Inc. Project 08CA44870, NC5756 for the Fire Protection Research Foundation January 12, 2009 Copyright 2009, Underwriters Laboratories Inc.

67 NC5756 Issued: 01/12/ CA44870 EXECUTIVE SUMMARY A series of two (2) large-scale fire tests were conducted to develop data regarding the level of protection provided by nominal K-14.0 gpm/psi ½ pendent ESFR ceiling sprinklers with an operating High Volume Low Speed (HVLS) 6-blade fan mounted 50-in below the ceiling. The sprinklers were installed to protect a double-row rack storage arrangement of cartoned, unexpanded Group A plastic test commodity to a nominal height of 20-ft under a 30-ft ceiling. The tests were conducted with 165 F temperature rated pendent ESFR ceiling sprinklers installed on 10 by 10-ft. spacing with the deflectors positioned 14-in. below the ceiling. Target arrays were positioned to both the north and south of the main array with 4-ft aisles. For both tests, the ceiling sprinkler system was supplied with water resulting in a nominal sprinkler discharge pressure of 50 psi. The first test was conducted with the main rack ignition scenario located between two sprinklers, and the center of 24-ft HVLS fan located offset from the main array 15-ft south of the ignition location. The test was conducted for 32 minutes (32:00), and a total of two (2) ceiling sprinklers operated during the test period. The fire in Test 1 was centered around the center bay above the ignition location and did not spread to either end of the main array or to the adjacent target arrays. The second test was conducted with the main rack ignition scenario located between four sprinklers, and the center of 24-ft HVLS fan positioned above ignition. The test was conducted for 31 minutes (31:00), and a total of eight (8) ceiling sprinklers operated during the test period. The fire in Test 2 was generally centered around the center two bays above the ignition location and did not spread to either end of the main array or to the adjacent target arrays. A summary of the parameters and results is presented in Table E1. 1

68 NC5756 Issued: 01/12/ CA44870 FIRE TEST DATE 10/08/ /10/2008 PARAMETERS Storage Type Double-Row Rack Double-Row Rack Commodity Type Cartoned, Unexpanded Group A Plastic Cartoned, Unexpanded Group A Plastic Nominal Storage Height (ft) Nominal Ceiling Height (ft) Nominal Clearance (ft) Aisle Width (in.) Ignition Location Between 2 Sprinklers Between 4 Sprinklers Sprinkler Temperature Rating ( F) Deflector to Ceiling (in) Nominal Sprinkler Discharge Coefficient K (gpm/psi ½ ) Nominal Sprinkler Discharge Pressure (psi) Sprinkler Spacing (ft x ft) 10 x x 10 Fan Size (ft) Fan Location 15-ft South of Ignition Above Ignition Fan Distance Below Ceiling (in) HVLS Fan Speed (rpm) RESULTS Length of Test (min:s) 32:00 31:00 First Ceiling Sprinkler Operation (min:s) 01:00 00:49 Last Ceiling Sprinkler Operation (min:s) 01:04 04:48 Number of Operated Ceiling Sprinklers 2 8 Peak Gas Temperature at Ceiling Above Ignition ( F) Maximum 1 Minute Average Gas Temperature at Ceiling Above Ignition ( F) Peak Steel Temperature at Ceiling Above Ignition ( F) Maximum 1 Minute Average Steel Temperature Above Ignition ( F) Fire Spread Across Aisle No No Fire Spread To the Ends of the Array No No Table E1 Summary of Test Parameters and Results 2

69 NC5756 Issued: 01/12/ CA44870 NOTE This Report was prepared as an account of a test sponsored by and under the direction of the Fire Protection Research Foundation. In no event shall UL be responsible for whatever use or nonuse is made of the information contained in this Report and in no event shall UL, its employees, or its agents incur any obligation or liability for damages arising out of or in connection with the use, or the inability to use, information contained in this Report. 3

70 NC5756 Issued: 01/12/ CA44870 Table Of Contents INTRODUCTION TEST FACILITY EQUIPMENT AND INSTRUMENTATION SPRINKLER SYSTEM INSTRUMENTATION 8 3 COMMODITY AND STORAGE ARRANGEMENT STANDARD GROUP A PLASTIC COMMODITY HVLS FAN ARRANGEMENT TEST 1 ARRANGEMENT Rack Storage Arrangement Test Ignition Method Test TEST 2 ARRANGEMENT Rack Storage Arrangement Test Ignition Method Test TEST PHOTOS AND RESULTS HVLS FAN PHOTOS TEST PHOTOS Test Test TEST RESULTS Test Test POST -TEST PHOTOS Test Test SUMMARY APPENDIX A... 1 GRAPHICAL PRESENTATION OF 1 GAS TEMPERATURES, STEEL TEMPERATURES, SYSTEM PRESSURE AND WATERFLOW - TEST 1 1 APPENDIX B... 1 GRAPHICAL PRESENTATION OF 1 GAS TEMPERATURES, STEEL TEMPERATURES, SYSTEM PRESSURE AND WATERFLOW - TEST 2 1 List of Tables TABLE 1 SPRINKLER WATER SUPPLY PARAMETERS... 7 TABLE 2 MOISTURE CONTENT OF GROUP A PLASTIC COMMODITY TABLE 3 SUMMARY OF TEST PARAMETERS AND RESULTS Table Of Figures FIGURE 1 TEST FACILITY... 6 FIGURE 2 NOMINAL K=14.0 PENDENT ESFR SPRINKLER... 7 FIGURE 3 SPRINKLER AND INSTRUMENTATION LOCATIONS... 9 FIGURE 4 STANDARD GROUP A PLASTIC COMMODITY FIGURE 5 TEST 1 RACK / FAN ARRANGEMENT PLAN VIEW

71 NC5756 Issued: 01/12/ CA44870 FIGURE 6 TEST 1 EAST / WEST ELEVATION VIEW FIGURE 7 TEST 1 IGNITION SCENARIO FIGURE 8 TEST 2 RACK / FAN ARRANGEMENT PLAN VIEW FIGURE 9 TEST 2 EAST / WEST ELEVATION VIEW FIGURE 10 TEST 2 IGNITION SCENARIO FIGURE 11 HVLS FAN INSTALLATION FIGURE 12 TEST 1 PHOTOS FIGURE 13 TEST 2 PHOTOS FIGURE 14 SPRINKLER OPERATIONS TEST FIGURE 15 DAMAGE ASSESSMENT PLAN VIEW TEST FIGURE 16 NORTH ROW DAMAGE ASSESSMENT TEST FIGURE 17 SOUTH ROW DAMAGE ASSESSMENT TEST FIGURE 18 SPRINKLER OPERATIONS TEST FIGURE 15 DAMAGE ASSESSMENT PLAN VIEW TEST FIGURE 19 NORTH ROW DAMAGE ASSESSMENT TEST FIGURE 20 SOUTH ROW DAMAGE ASSESSMENT TEST FIGURE 21 POST TEST PHOTOS TEST FIGURE 22 POST TEST PHOTOS TEST

72 NC5756 Issued: 01/12/ CA44870 INTRODUCTION This Test Report describes the Special Service Investigation conducted for the Fire Protection Research Foundation to develop fire test data relative to the influence of High Volume Low Speed (HVLS) fans on the level of protection provided for a fire scenario involving rack storage of cartoned, unexpanded Group A plastic commodity protected by Early Suppression Fast Response (ESFR) sprinklers. 1 TEST FACILITY The fire tests were conducted in Underwriters Laboratories Inc. s large-scale fire test facility located in Northbrook, Illinois. The large-scale fire test building used for this investigation includes four fire test areas that are used to develop data on the fire growth and fire suppression characteristics of commodities, as well as the fire suppression characteristics of automatic water sprinkler systems. A schematic of the test facility is shown in Figure 1. ADD Test Facility Large Scale Fire Test Facility Heat Release Calorimeter & RDD PDPA Test Facility Conditioning Room Warehouse Figure 1 Test Facility The fire tests were conducted in a 120 by 120 by 54-ft. high room fitted with a 100 x 100-ft adjustable height, smooth, flat, horizontal ceiling. The ceiling was positioned 30-ft above the floor. The test room was equipped with an exhaust system including a regenerative thermal oxidizing (RTO) smoke abatement system. Ambient temperature outside make-up air at approximately 60,000 cfm was provided through four inlet ducts positioned along the walls of the test facility after first sprinkler operation. Prior to first sprinkler operation, ambient temperature outside make-up air was drawn in through the ducts at approximately 30,000 cfm. The floor of the test facility was smooth, flat and surrounded with a grated drainage trench to insure adequate drainage from the test area. The water runoff from the suppression system drain was collected through a 180,000-gallon water treatment system. 6

73 NC5756 Issued: 01/12/ CA EQUIPMENT AND INSTRUMENTATION 2.1 Sprinkler System One hundred (100) pendent ESFR sprinklers having a 165 F temperature rating and a nominal discharge coefficient (K) of 14.0 gpm/psi 1/2 were installed on 10 x 10-ft. spacing in a closed head, wet pipe, automatic sprinkler system at the ceiling level. The sprinklers were supplied through a looped piping system consisting of 2 ½-in. diameter branch lines. The sprinklers were positioned with the distance between the deflector and ceiling measured to be 14-in. The water supply to the ceiling sprinklers was controlled to achieve the parameters shown in Table 1. Close-up photos of the sprinklers used in this test are shown in Figure 2. A plan view of the sprinkler and instrumentation layout is shown in Figure 3. Test No. Sprinkler Location Nominal flow per sprinkler (gpm) Nominal pressure at sprinkler (psig) 1 Ceiling Ceiling Table 1 Sprinkler Water Supply Parameters Figure 2 Nominal K=14.0 Pendent ESFR Sprinkler 7

74 NC5756 Issued: 01/12/ CA Instrumentation The instrumentation used in the testing consisted of the following devices: One hundred (100) 1/16-in. diameter, Type K inconel sheathed thermocouples located below the ceiling adjacent to each sprinkler Three 1/16-in. diameter, Type K inconel sheathed thermocouple located 6, 12 and 18-in. below the ceiling above the ignition location Five 1/16-in. diameter, Type K inconel sheathed thermocouples embedded in a 50.5-in. long steel angle attached to the bottom of the ceiling directly above the fire A pressure transducer in a psi range was used to measure the water pressure in the sprinkler system A 12-in. magnetic flow meter in the gpm range was used to measure the water flow rate Stopwatches and timing devices located within the data acquisition system will be used to monitor and record significant events during the fire test Moisture meter to measure the commodity moisture content of the fuel package Video and Infrared cameras were used to capture and record images Electronic data acquisition system with a one-second-scan rate to obtain the data generated 8

75 NC5756 Issued: 01/12/ CA44870 Figure 3 Sprinkler and Instrumentation Locations 9

76 NC5756 Issued: 01/12/ CA COMMODITY AND STORAGE ARRANGEMENT 3.1 Standard Group A Plastic Commodity The standard Group A Plastic commodity was used in both the main and center bays of the target arrays and consisted of rigid crystalline polystyrene cups (empty, 16 oz size) packaged in compartmented, single-wall, corrugated cardboard cartons. Cups are arranged in five layers, 25 per layer for a total of 125 per carton. The compartmentalization was accomplished with single wall corrugated cardboard sheets to separate the five layers and vertical interlocking single-wall corrugated cardboard dividers to separate the five rows and five columns of each layer. Eight 21-in. cube cartons, arranged 2x2x2, comprise a pallet load. Each pallet load is supported by a two-way, 42 in. by 42 in. by 5 in., slatted deck hardwood pallet. Figure 4 Standard Group A Plastic Commodity Both box and pallet samples from the test commodity were taken to determine the moisture content. The moisture content for the commodity used is presented in Table 2. Test No. Item Average Moisture Content Maximum Moisture Content Box 6.5% 7.0% 1 Pallet 7.3% 8.3% 2 Box 4.7% 5.5% Pallet 7.9% 8.5% Table 2 Moisture Content of Group A Plastic Commodity 3.2 HVLS Fan Arrangement A High Volume Low Speed (HVLS) 6-blade fan with a 24-ft diameter was installed in the test cell, with the fan hub mounted 50-in below the ceiling for both tests. The fan operated such that the air movement was in the downward direction, and the variable speed motor drive was arranged for the fan to operate at maximum speed (63 RPM). 10

77 NC5756 Issued: 01/12/ CA44870 For Test 1, the center of the fan was positioned 15-ft south of the ignition location. For Test 2, the fan was positioned above ignition. 3.3 Test 1 Arrangement RACK STORAGE ARRANGEMENT TEST 1 The main array consisted of industrial racks utilizing steel upright and steel beam construction. The nominal 16-ft high by 32-in. wide rack members were arranged to provide a double-row main rack with four 96-in. bays and four tiers in each row. Beam tops were positioned in the racks at vertical tier heights in 10, 70, 130 and 190 in. above the floor. The geometric center of the double row rack was positioned under two sprinklers in the test room, as shown in Figure 5. The target arrays consisted of industrial, single-row rack utilizing steel upright and steel beam construction. The nominal 16-ft high by 32-in. wide rack system was arranged in a single-row target rack with three 96-in. bays with four tiers in each row. Cartoned, unexpanded Group A plastic commodity was used in the center two bays of each target and Standard Class II commodity was used in the outside bays of each target. The commodities were positioned at 10, 70, 130, and 190-in. above the floor. The target rack was positioned across a 4-ft aisle space from the main array on the north and south sides. The bays of the main and target racks were loaded with the standard test commodity as described to provide nominal 6 in. wide longitudinal and transverse flue spaces throughout the test array. The HVLS fan was positioned with the center of 24-ft HVLS fan located above the main array 15-ft south of the ignition location, with the fan hub positioned 50-in below the ceiling as shown in Figure 6. 11

78 NC5756 Issued: 01/12/ CA44870 Figure 5 Test 1 Rack / Fan Arrangement Plan View 12

79 NC5756 Issued: 01/12/ CA44870 Figure IGNITION METHOD TEST 1 Test 1 East / West Elevation View Ignition was accomplished by using four half-standard cellulose cotton igniters. The igniters were constructed from 3 in. by 3 in. long cellulosic bundles each soaked with 4 oz. of gasoline and wrapped in a polyethylene bags. The igniters were positioned 10-in off the floor and located in the longitudinal flue at the center of the main array, as shown in Figure 7. 13

80 NC5756 Issued: 01/12/ CA44870 Figure 7 Test 1 Ignition Scenario 3.4 Test 2 Arrangement RACK STORAGE ARRANGEMENT TEST 2 The main array consisted of industrial racks utilizing steel upright and steel beam construction. The nominal 16-ft high by 32-in. wide rack members were arranged to provide a double-row main rack with four 96-in. bays and four tiers in each row. Beam tops were positioned in the racks at vertical tier heights in 10, 70, 130 and 190 in. above the floor. The geometric center of the double row rack was positioned between four sprinklers in the test room, as shown in Figure 8. The target arrays consisted of industrial, single-row rack utilizing steel upright and steel beam construction. The nominal 16-ft high by 32-in. wide rack system was arranged in a single-row target rack with three 96-in. bays with four tiers in each row. Cartoned, unexpanded Group A plastic commodity was used in the center two bays of each target and Standard Class II commodity was used in the outside bays of each target. The commodities were positioned at 10, 70, 130, and 190-in. above the floor. The target rack was positioned across a 4-ft aisle space from the main array on the north and south sides. 14

81 NC5756 Issued: 01/12/ CA44870 The bays of the main and target racks were loaded with the standard test commodity as described to provide nominal 6 in. wide longitudinal and transverse flue spaces throughout the test array. The HVLS fan was positioned with the center of 24-ft HVLS fan located above the ignition location in the main array, with the fan hub positioned 50-in below the ceiling as shown in Figure 9. 15

82 NC5756 Issued: 01/12/ CA44870 Figure 8 Test 2 Rack / Fan Arrangement Plan View 16

83 NC5756 Issued: 01/12/ CA44870 Figure 9 Test 2 East / West Elevation View IGNITION METHOD TEST 2 Ignition was accomplished by using four half-standard cellulose cotton igniters. The igniters were constructed from 3 in. by 3 in. long cellulosic bundles each soaked with 4 oz. of gasoline and wrapped in a polyethylene bags. The igniters were positioned 10-in off the floor and located in the longitudinal flue at the center of the main array, as shown in Figure

84 NC5756 Issued: 01/12/ CA44870 Figure 10 Test 2 Ignition Scenario 4 TEST PHOTOS AND RESULTS 4.1 HVLS Fan Photos Fan Mounting Figure 11 Fan Hub (shown 50-in from Ceiling) HVLS Fan Installation 18

85 NC5756 Issued: 01/12/ CA Test Photos TEST 1 View from NE View from NW View from SE View from East View from West Igniters 19

86 NC5756 Issued: 01/12/ CA44870 View from East (prior to ignition) View from East (after sprinkler operation) TEST 2 Figure 12 Test 1 Photos View from NW View from NE View from SE View from North 20

87 NC5756 Issued: 01/12/ CA44870 Fan Detail From East Igniters View from East (prior to ignition) Figure 13 Test 2 Photos View from East (after sprinkler operation) 4.3 Test Results TEST 1 The test was conducted for 32 minutes (32:00). A total of two (2) ceiling sprinklers operated during the 32-minute test period. The first sprinkler activated at one minute zero seconds (01:00) after ignition and the last sprinkler operated at one minute four seconds (01:04) after ignition. The fire in Test 1 was generally contained within the center bay above the ignition location and did not spread to either end of the main array or to the adjacent targets. The maximum oneminute average steel beam temperature measured above ignition was 106ºF and the maximum one-minute average gas temperature measured above ignition was 131ºF. The HVLS fan remained operational throughout and following the test. 21

88 NC5756 Issued: 01/12/ CA44870 Sprinkler operation times are presented in Figure 14, and plan / elevation views of the extent of fire are presented in Figures 15 through 17. Post-test photos are presented in Figure 22. Tabulated test results are presented in Table 3. Temperature data vs. time charts are presented in Appendix A TEST 2 The second large-scale fire test was conducted for 31 minutes (31:00). A total of eight (8) ceiling sprinklers operated during the 31-minute test period. The first sprinkler activated at forty-nine seconds (00:49) after ignition and the last sprinkler operated at four minutes fortyeight seconds (04:48) after ignition. The fire in Test 2 was generally contained within the center two bays above the ignition location and did not spread to either end of the main array or to the adjacent targets. The maximum one-minute average steel beam temperature measured above ignition was 142ºF and the maximum one-minute average gas temperature measured above ignition was 201ºF. The HVLS fan remained operational throughout and following the test. Sprinkler operation times are presented in Figure 18, and plan / elevation views of the extent of fire are presented in Figures 19 through 21. Post-test photos are presented in Figure 23. Tabulated test results are presented in Table 3. Temperature data vs. time charts are presented in Appendix B. 22

89 NC5756 Issued: 01/12/ CA44870 FIRE TEST DATE 10/08/ /10/2008 PARAMETERS Storage Type Double-Row Rack Double-Row Rack Commodity Type Cartoned, Unexpanded Group A Plastic Cartoned, Unexpanded Group A Plastic Nominal Storage Height (ft) Nominal Ceiling Height (ft) Nominal Clearance (ft) Aisle Width (in.) Ignition Location Between 2 Sprinklers Between 4 Sprinklers Sprinkler Temperature Rating ( F) Deflector to Ceiling (in) Nominal Sprinkler Discharge Coefficient K (gpm/psi ½ ) Nominal Sprinkler Discharge Pressure (psi) Sprinkler Spacing (ft x ft) 10 x x 10 Fan Size (ft) Fan Location 15-ft South of Ignition Above Ignition Fan Distance Below Ce iling (in) HVLS Fan Speed (rpm) RESULTS Length of Test (min:s) 32:00 31:00 First Ceiling Sprinkler Operation (min:s) 01:00 00:49 Last Ceiling Sprinkler Operation (min:s) 01:04 04:48 Number of Operated Ceiling Sprinklers 2 8 Peak Gas Temperature at Ceiling Above Ignition ( F) Maximum 1 Minute Average Gas Temperature at Ceiling Above Ignition ( F) Peak Steel Temperature at Ceiling Above Ignition ( F) Maximum 1 Minute Average Steel Temperature Above Ignition ( F) Fire Spread Across Aisle No No Fire Spread To the Ends of the Array No No Table 3 Summary of Test Parameters and Results 23

90 NC5756 Issued: 01/12/ CA44870 Figure 14 Sprinkler Operations Test 1 24

91 NC5756 Issued: 01/12/ CA44870 Figure 15 Damage Assessment Plan View Test 1 25

92 NC5756 Issued: 01/12/ CA44870 Figure 16 North Row Damage Assessment Test 1 Figure 17 South Row Damage Assessment Test 1 26

93 NC5756 Issued: 01/12/ CA44870 Figure 18 Sprinkler Operations Test 2 27

94 NC5756 Issued: 01/12/ CA44870 Figure 19 Damage Assessment Plan View Test 2 28

95 NC5756 Issued: 01/12/ CA44870 Figure 20 North Row Damage Assessment Test 2 Figure 21 South Row Damage Assessment Test 2 29

96 NC5756 Issued: 01/12/ CA Post-Test Photos TEST 1 View from SW View from NW View from South View from SE Ignition Location View above Ignition Figure 22 Post Test Photos Test 1 30

97 NC5756 Issued: 01/12/ CA TEST 2 View from NE View from SE View from South View from North View from East View from SW Figure 23 Post Test Photos Test 2 31

98 NC5756 Issued: 01/12/ CA SUMMARY A series of two (2) large-scale fire tests were conducted to develop data regarding the level of protection provided by nominal K-14.0 gpm/psi ½ pendent ESFR ceiling sprinklers with an operating High Volume Low Speed (HVLS) 6-blade fan mounted 50-in below the ceiling. The sprinklers were installed to protect a double-row rack storage arrangement of cartoned, unexpanded Group A plastic test commodity to a nominal height of 20-ft under a 30-ft ceiling. The tests were conducted with 165 F temperature rated pendent ESFR ceiling sprinklers installed on 10 by 10-ft. spacing with the deflectors positioned 14-in. below the ceiling. Target arrays were positioned to both the north and south of the main array with 4-ft aisles. For both tests, the ceiling sprinkler system was supplied with water resulting in a nominal sprinkler discharge pressure of 50 psi. The first test was conducted with the main rack ignition scenario located between two sprinklers, and the center of 24-ft HVLS fan located offset from the main array 15-ft south of the ignition location. The test was conducted for 32 minutes (32:00), and a total of two (2) ceiling sprinklers operated during the test period. The first sprinkler activated at one minute zero seconds (01:00) after ignition and the last sprinkler operated at one minute four seconds (01:04) after ignition. The maximum one-minute average steel beam temperature measured above ignition was 106ºF and the maximum one-minute average gas temperature measured above ignition was 131ºF. The fire in Test 1 was centered around the center bay above the ignition location and did not spread to either end of the main array or to the adjacent target arrays. The second test was conducted with the main rack ignition scenario located between four sprinklers, and the center of 24-ft HVLS fan positioned above ignition. The test was conducted for 31 minutes (31:00), and a total of eight (8) ceiling sprinklers operated during the test period. The first sprinkler activated at forty-nine seconds (00:49) after ignition and the last sprinkler operated at four minutes forty-eight seconds (04:48) after ignition. The maximum one-minute average steel beam temperature measured above ignition was 142ºF and the maximum oneminute average gas temperature measured above ignition was 201ºF. The fire in Test 2 was generally centered around the center two bays above the ignition location and did not spread to either end of the main array or to the adjacent target arrays. A summary of the parameters and results is presented in Table 3. Report by: Reviewed by: Christopher Gates Michael McCormick Project Engineer Staff Engineering Associate Tel: Tel: Fax: Fax:

99 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 APPENDIX A Graphical Presentation of Gas Temperatures, Steel Temperatures, System Pressure and Waterflow - Test 1 Index: Sprinkler Temperatures System Pressure System Waterflow Steel Temperatures above Ignition Air Temperatures above Ignition Page A2 A11 A12 A12 A13 A13 A-1

100 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.001 Spk.002 Spk.003 Spk.004 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.006 Spk.007 Spk.008 Spk.009 Spk Time (Min) A-2

101 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.011 Spk.012 Spk.013 Spk.014 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.016 Spk.017 Spk.018 Spk.019 Spk Time (Min) A-3

102 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.021 Spk.022 Spk.023 Spk.024 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.026 Spk.027 Spk.028 Spk.029 Spk Time (Min) A-4

103 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.031 Spk.032 Spk.033 Spk.034 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.036 Spk.037 Spk.038 Spk.039 Spk Time (Min) A-5

104 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.041 Spk.042 Spk.043 Spk.044 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.046 Spk.047 Spk.048 Spk.049 Spk Time (Min) A-6

105 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.051 Spk.052 Spk.053 Spk.054 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.056 Spk.057 Spk.058 Spk.059 Spk Time (Min) A-7

106 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.061 Spk.062 Spk.063 Spk.064 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.066 Spk.067 Spk.068 Spk.069 Spk Time (Min) A-8

107 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.071 Spk.072 Spk.073 Spk.074 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.076 Spk.077 Spk.078 Spk.079 Spk Time (Min) A-9

108 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.081 Spk.082 Spk.083 Spk.084 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.086 Spk.087 Spk.088 Spk.089 Spk Time (Min) A-10

109 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.091 Spk.092 Spk.093 Spk.094 Spk Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.096 Spk.097 Spk.098 Spk.099 Spk Time (Min) A-11

110 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Pressure (PSI) 30 Inner Loop Pressure Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler System Flow (GPM) Flow Time (Min) A-12

111 NC5756 Appendix A Test 1 Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Steel Temperatures above Ignition (Deg C) TC 1 TC 2 TC 3 TC 4 TC Time (Min) Fire Protection Research Foundation HVLS Test 1 - Nominal 20-ft. Group A Commodity on Racks Between Two K14 ESFR Sprinklers under a 30-ft. Ceiling Air Temperatures above Ignition (Deg C) in. B/C 12-in. B/C 18-in. B/C Time (Min) A-13

112 NC5756 Appendix B Graphs Issued 01/12/ CA44870 APPENDIX B Graphical Presentation of Gas Temperatures, Steel Temperatures, System Pressure and Waterflow - Test 2 Index: Sprinkler Temperatures System Pressure System Waterflow Steel Temperatures above Ignition Air Temperatures above Ignition Page B2 B11 B12 B12 B13 B13 B-1

113 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.001 Spk.002 Spk.003 Spk.004 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.006 Spk.007 Spk.008 Spk.009 Spk Time (Min) B-2

114 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.011 Spk.012 Spk.013 Spk.014 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.016 Spk.017 Spk.018 Spk.019 Spk Time (Min) B-3

115 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.021 Spk.022 Spk.023 Spk.024 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.026 Spk.027 Spk.028 Spk.029 Spk Time (Min) B-4

116 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.031 Spk.032 Spk.033 Spk.034 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.036 Spk.037 Spk.038 Spk.039 Spk Time (Min) B-5

117 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.041 Spk.042 Spk.043 Spk.044 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.046 Spk.047 Spk.048 Spk.049 Spk Time (Min) B-6

118 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.051 Spk.052 Spk.053 Spk.054 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.056 Spk.057 Spk.058 Spk.059 Spk Time (Min) B-7

119 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.061 Spk.062 Spk.063 Spk.064 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.066 Spk.067 Spk.068 Spk.069 Spk Time (Min) B-8

120 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.071 Spk.072 Spk.073 Spk.074 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.076 Spk.077 Spk.078 Spk.079 Spk Time (Min) B-9

121 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures 45 Note: Data for Spk. 084 not obtained (Deg C) Spk.081 Spk.082 Spk.083 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.086 Spk.087 Spk.088 Spk.089 Spk Time (Min) B-10

122 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.091 Spk.092 Spk.093 Spk.094 Spk Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Temperatures (Deg C) Spk.096 Spk.097 Spk.098 Spk.099 Spk Time (Min) B-11

123 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler Pressure (PSI) 30 Inner Loop Pressure Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Sprinkler System Flow (GPM) 600 Flow Time (Min) B-12

124 NC5756 Appendix B Graphs Issued 01/12/ CA44870 Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Steel Temperatures above Ignition (Deg C) TC 1 TC 2 TC 3 TC 4 TC Time (Min) Fire Protection Research Foundation HVLS Test 2 - Nominal 20-ft. Group A Commodity on Racks Between Four K14 ESFR Sprinklers under a 30-ft. Ceiling Air Temperatures above Ignition (Deg C) 60 6-in. B/C 12-in. B/C 18-in. B/C Time (Min) B-13

125 APPENDIX C REQUEST FOR PROPOSALS PUBLISHED BY FPRF

126 BATTERYMARCH PARK, QUINCY, MASSACHUSETTS, U.S.A REQUEST FOR PROPOSALS - PROJECT MANAGEMENT HVLS FANS AND SPRINKLER OPERATION PHASE I RESEARCH PROGRAM BACKGROUND: High volume low speed (HVLS) fans are in increasing use in storage and manufacturing facilities. However, the interaction of these fans and automatic sprinkler operation is unknown. The Foundation is initiating a research program to investigate this issue, with a goal to determine appropriate spacing and other installation requirements for these fans in the presence of automatic sprinklers. The attached research plan was developed by Southwest Research Institute at the request of the Foundation. It lays out a comprehensive research plan as well as a short term Phase I plan designed to focus on two specific high priority performance issues. The Foundation is seeking a project manager to oversee the Phase I research program. PHASE I RESEARCH OBJECTIVE: To explore the impact of HVLS fans on ESFR sprinkler performance for rack and palletized storage of commodities. Required fan shut off time (and detection means), and fan blade impact on sprinkler spray patterns will be the focus of the Phase I program. SCOPE OF PROJECT MANAGEMENT SERVICES: 1. Literature Review gather and review available information on fan performance and benchmark ESFR sprinkler tests. 2. Finalization of Detailed Research Plan using this review and the SWRI plan as a starting point, develop a targeted and prioritized research plan to include sequencing and number of tests, configurations, etc to be undertaken with funds available. A goal will be to have some preliminary information available for the NFPA 13 Sprinkler Installation Criteria Technical Committee meeting in September of this year. 3. Coordination of Testing coordination with one or more fire test laboratories to implement the test program. 4. Analysis of Test Results and Reporting review and analysis of test data, status reporting to the NFPA 13 Committee, and development of a detailed Phase II plan of additional testing/modeling or research needed to meet the overall objective of the research program.