Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Der Artikel untersucht die Leistungsfähigkeit von bedarfsgeregelten, klappenoptimierten Lüftungssystemen (DCV) bei Brandereignissen, wobei der Schwerpunkt auf den potenziellen Versagensmechanismen und ihren Auswirkungen auf die Rauch- und Druckkontrolle liegt. Sie präsentiert Ergebnisse von umfassenden Brandversuchen, die in einem simulierten Schulgebäude durchgeführt wurden, zu dem ein Klassenzimmer, ein Büro und ein Flur gehörten. Bei den Tests wurden verschiedene Brandszenarien mit unterschiedlichen Brennstoffen wie E-Rollern, Polyurethanschaum-Matratzen und Propangas untersucht, um das Lüftungssystem herauszufordern und Versagensmechanismen zu identifizieren. Die Studie zeigt, dass DCV-Dämpfer ohne Brandschutz aufgrund von Hitzeeinwirkung versagen können, was zu unbeabsichtigtem Schließen oder Öffnen der Dämpfer führt und die Lüftungsbilanz beeinträchtigt. Der Ausfall eines einzigen Dämpfers kann mehrere Räume betreffen und die Evakuierungsbedingungen beeinträchtigen. Der Artikel behandelt auch die allmähliche Verstopfung von Absaugfiltern und ihre Auswirkungen auf die Leistung des Lüftungssystems. Sie kommt zu dem Schluss, dass Konstruktionsänderungen und zusätzliche Sicherheitsmaßnahmen wie Bypasssysteme notwendig sind, um die Feuerbeständigkeit von klappenoptimierten Lüftungssystemen zu gewährleisten. Die Ergebnisse liefern wertvolle Erkenntnisse über das Verhalten moderner Lüftungssysteme bei Brandereignissen und geben Empfehlungen zur Verbesserung der Brandschutzstrategien in Gebäuden.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
Modern heating, ventilation, and air conditioning (HVAC) systems are complex, interconnected systems optimised to be energy efficient. Damper-optimised demand-controlled ventilation systems (DCV) minimise energy consumption by using a dedicated control unit that calculates the optimal fan speed based on room sensors and the feedback from all DCV dampers, which each measures the airflow rate and adjusts its damper angle accordingly. In buildings that do not use a compartmentation strategy in the event of a fire, it is crucial that the ventilation system is pressurised and provides balanced ventilation in order to prevent smoke from spreading via the ventilation system and to avoid creating pressure imbalances, which may impair evacuation. In the present study, two full-scale fire tests from a series of 14 tests in a mock-up building equipped with a damper-optimised DCV system are presented, and the ventilation system’s performance during the fire is assessed. The tests revealed various failure mechanisms caused by heat exposure, leading to individual damper uncontrolled opening or closing or the building management system losing contact with all dampers. Furthermore, it was shown that the failure of individual dampers and the gradual clogging of the extraction filter can affect the pressure balance in other parts of the building outside the fire room and increase the risk of smoke spreading through the ventilation ducts.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
HVAC (heating, ventilation, and air conditioning) systems with accompanying control in buildings have changed a lot during the last 20 years, from simple systems with constant airflow to today’s more complex demand-controlled systems with VAV (variable air volume)- or DCV (demand-controlled ventilation)-dampers that regulate the air-flow after the use of the building. The aim of regulating the airflow is to enhance the air quality and decrease the energy use in the buildings. VAV comprises solutions where a true occupancy detector controls the damper position to preset positions but where there is no feedback on the effect of the position. Whereas DCV dampers automatically regulate after demand measured at room level, e.g. from combined temperature and carbon dioxide (\(\hbox {CO}_2\))-sensors. The most energy-efficient but also most complex DCV system is a so-called damper-optimised system, in which the fan speed in the air handling unit (AHU) is regulated by a dedicated control unit that takes the required airflow rate, the supplied airflow rate as well as the damper angle for all dampers in the system into account.[1].
In the event of a fire, the ventilation ducts present a potential path for fire and smoke spreading [2, 3]. Two different strategies can be employed to prevent the spread of fire and smoke in ventilation systems: compartmentation and extraction. The compartmentation strategy involves installing fire dampers in fire compartment walls and ceilings that will close both the extraction and supply air ducts in case of fire. This ensures the fire is kept inside the room of origin. However, a high over-pressure can be expected, resulting in smoke leakages through weak points in the construction [4]. The extraction strategy keeps the fans in operation during a fire, venting out the smoke from the fire room and, most importantly, pressurising the ventilation ducts, which forces airflow in the desired direction and prevents smoke from entering the supply air ducts.
Anzeige
During a fire, pressure differences will appear between the fire room and the surrounding parts of the building. This has consequences for the spread of fire and smoke, evacuation conditions, and damage to the construction. At the same time, there is a risk of damage to security systems, technical equipment, and other installations [5]. Indoor fires are characterised by a natural course of overpressure, pressure equalisation and underpressure. In an early phase, often referred to as the fire growth phase, an overpressure will quickly build up in the fire room caused by the room functioning as a physical barrier that resists the expanding hot air and fire gases from the combustion process. This phase lasts until balance is achieved between the heat gain from the fire and the heat emitted from the room. The pressure will then gradually decrease due to a lack of oxygen or a lack of combustible material, and an underpressure in the fire room is established. At last, the pressure increases again to return to its original level before ignition. The precise pressure development depends on the air tightness, heat losses, and fire development[6]. The pressure development during the fire growth phase is of particular interest since new research shows that the overpressure can reach critical values in a short time. Studies show that the pressure levels can be so high that escape route doors can be hard to open manually, smoke spread can arise quickly and in unintentional ways, and elements in the building construction may collapse [3‐5, 7]. For a functioning extraction strategy that prevents fire and smoke from spreading through the ventilation ducts, it is, therefore, crucial that the ventilation system is able to provide balanced ventilation and sufficient high pressure throughout the entire fire event.
The extraction strategy is commonly used in Norway and has been so for many years [8]. However, the performance of modern damper-optimised DCV systems during a fire has not yet been investigated, which is of special concern since currently used VAV- and DCV-dampers are not fire-rated and only designed to function at operating temperature levels, typically in the range of \(50 \, ^\circ {\hbox {C}}\). Furthermore, the question of how the entire ventilation system reacts to the potential failure of individual components, such as a single DCV damper, has not yet been answered for highly interconnected systems such as a damper-optimised DCV system.
The present work presents the results from two full-scale fire tests out of a series of 14 tests that were conducted in a mock-up school building. The main objectives of the test campaign were
to investigate how central components of a ventilation system, such as DCV dampers and extraction filters, react to smoke (i.e., heat and soot) exposure,
and to evaluate how failures of individual components affect the ventilation system’s performance in terms of smoke and pressure control.
As of today, the central components of a damper-optimised ventilation system are not fire-rated, and several hypothetical failure mechanisms, such as exceeding the temperature range of the dampers flow measurement units, melting of plastic components, and clogging of measuring rods or filters, were identified prior to the experiments. The experimental campaign was designed to challenge these failure mechanisms by involving different types of fuels, fire locations and ventilation scenarios. The fuels consisted of fire in an e-scooter, polyurethane (PU) foam mattress, propane gas, and a combination of gas and PU foam. Burning e-scooters (Li-ion battery fires) produce, for example, smaller particles (containing carbon, oxygen, and heavy metals) compared to PU fires (containing mostly organic particles). On the other hand, gas fires are well suited to control the heat release rate precisely.
For all identified failure mechanisms, the following two questions were addressed: Can the fault cause a pressure imbalance between rooms, especially other rooms than the fire room, and can the fault cause smoke to spread via the ventilation ducts?
Anzeige
The present study focuses on the results from two of the tests, which cover all identified failure mechanisms and their consequences. However, other tests are mentioned where necessary to discuss the repeatability of the results. A description of the entire test campaign, in Norwegian only, can also be found in [9, 10].
2 Experimental Setup
The experimental setup and the test scenarios are described in this section. The experimental setup is also partly described by Meraner et al. [11], who used the data from one of the tests from the presented experimental campaign as the basis for a computational fluid dynamic simulation, including a model of the damper-optimised demand-controlled ventilation system during a fire.
2.1 Mock-Up Building
The mock-up building where the tests were performed was designed to model a school. The building consisted of three rooms: a classroom, a corridor, and an office, with a total area of 101 \(\hbox {m}^2\). The building walls consisted of one layer of 12 mm thick oriented strand boards (OSB) with one layer of 12.5 mm thick gypsum boards mounted on the inside. No insulation or exterior cladding was mounted on the walls. EI30 fire doors were used both as external and internal doors. The roof consisted of 12 mm thick OSB boards. Above the roof, a layer of 120 mm thick stone wool insulation (Rockwool FLEXI A-PLATE TRD) was added in the corridor, and a layer of 95 mm thick insulation was added in the corridor and in the classroom. Corrugated steel sheets were used as the top layer of the roof. Indoors, the ceiling was protected with one layer of 30 mm thick stone wool insulation plates (SeaRox). The dimensions of the rooms can be seen in Fig. 1, together with the installed ventilation system. The ceiling height varied from 3.01 m in the corridor to 3.18 m in the classroom, because the building was built on a sloped concrete surface. The classroom and the office did not share any internal walls in order to be able to distinguish any potential smoke spreading through the ventilation system and smoke spreading through leakage in the room envelope.
Fig. 1
Mock-up building and ventilation system layout. Supply ductwork is shaded yellow, and extraction ductwork is shaded blue (Color figure online)
In order to evaluate specific leak areas and possible leak paths for the different rooms, individual blower door tests according to ISO 9972 [12] were conducted for each room, in which the leakage volume flow rate, \(V_L\) in \({\hbox {m}}^{3}{\hbox {s}}^{-1}\), for a pressure difference, \(\Delta P\) in Pa, across the blower door, is measured. The volume flow is calculated based on the equation below:
$$\begin{aligned} \dot{V}_L = C_d A_{L,ref} \left( \frac{\left| \Delta P \right| }{\Delta P_{ref}} \right) ^{n_P-0.5} \text {sign}\left( \Delta P \right) \sqrt{2\frac{\left| \Delta P \right| }{\rho }}, \end{aligned}$$
(1)
where \(C_d\) is a discharge coefficient (typically assumed to be 1 for a leakage test), \(\rho \) is the ambient air density in \({\hbox {kg}}~{\hbox {m}}^{-3}\), \(A_{L,ref}\) is reference leakage area in \({\hbox {m}}^{2}\), \(\Delta P_{ref}\) is the reference pressure (typically 50 Pa), and \(n_p\) is the pressure exponent (the leakage area typically increases with \(\Delta P\)). Both overpressure and underpressure tests were performed at pressure differences of nominally 10 to 60 Pa in steps of 10 Pa. A least squares fit was employed to determine \(A_{L,ref}\) and \(n_p\) for that configuration based on the results from the different pressure differences. For the corridor, two sets of measurements were conducted, one with the doors to the adjacent rooms sealed with tape and one with the doors as built, making it possible to establish a representative leakage area for a single door. The measured leakage areas for the different rooms are found in Table 1. The average specific effective leakage area (envelope) at 50 Pa for the rooms in the mock-up building was 0.4 \({\hbox {m}}^{2}~{\hbox {m}}^{-2}\), which is considerably smaller than the reported values of 1.3 \({\hbox {m}}^{2}~{\hbox {m}}^{-2}\) and 3.7 \({\hbox {m}}^{2}~{\hbox {m}}^{-2}\) from two different actual Norwegian school buildings [9]. Building the mock-up building relatively tight was done to challenge the ventilation strategy on regulating the pressure differences and maintaining an acceptable door environment, allowing people to open and close the doors. This is because more air tight compartments will cause a larger pressure increase during the fire growth phase compared to compartments with a larger leakage area.
Table 1
Blower door test configuration and effective leak area (absolute and normalised by the room envelope) measurements
Room
Configuration (50 Pa)
Leak area
(\({\hbox {cm}}^{2}\))
Specific effective leak area
(\({\hbox {cm}}^{2}~{\hbox {m}}^{-2}\))
Flow rate at 50 Pa
(\({\hbox {m}}^{3}~{\hbox {h}}^{-1}\))
Classroom
Overpressure
56 (±9.7 %)
0.271
185
Underpressure
52 (±9.7 %)
0.251
172
Corridor
Overpressure
96 (±3.2 %)
0.563
316
Underpressure
79 (±0.6 %)
0.464
260
Corridor (sealed doors)
Overpressure
84 (±9.0 %)
0.491
276
Underpressure
75 (±2.5 %)
0.440
247
Office
Overpressure
31 (±7.1 %)
0.477
101
Underpressure
24 (±7.1 %)
0.378
80
2.2 Demand-Controlled Ventilation System
A damper-optimised balanced demand-controlled ventilation (DVC) system was installed in the mock-up building. During normal operation, the airflow rates are regulated based on the input from the pressure-independent supply and extraction variable air volume (VAV) dampers at room level, active \(\hbox {CO}_2\) sensors and passive temperature sensors communicating with the programmable logic controller under station (PLC-US), which determined the required airflow rates continuously.
The installed AHU of type Exhausto Topp-4-R-NA2-3-1 was able to deliver an airflow rate of 4 500\({\hbox {m}}^{3}~{\hbox {h}}^{-1}\) (250 Pa external pressure), corresponding to 70 % of maximum nominal capacity. The AHU had two fans, one supply and one extraction fan, a regenerative heat exchanger (RHE) and a heating battery; however, the heating battery was not active during the tests. Four ePM1 60 % (IF83: 592\(\times \)442\(\times \)535/10) bag filters were installed in the AHU, two on the extraction side and two on the supply side. The filters on the extraction side were replaced between each test. The maximum operating temperature of the filters is \(70~^\circ {\hbox {C}}\).
Multi-criteria smoke detectors of type AutroGuard\(\circledR \)V-430 were mounted in each room and connected to the building management system (BMS) via the fire alarm control panel. In the case of fire detection, the BMS system responds to provide \(\dot{V}_{max}\) to all rooms, by increasing the AHU power to obtain 100 % of the designed maximal airflow rate for the building. Depending on the system, this does not necessarily require fan speed to operate at 100 %. The optimizer function of the PLC-US adjusts the DCV damper positions to supply \(\dot{V}_{max}\) to the compartments. As a once-through system, providing 100 % supply means also having 100 % extraction, which means the RHE will provide maximum heat exchange between the extraction and supply air streams.
In addition to the rooms in the mock-up building, the ventilation system was designed to cover an additional area of 300 \(\hbox {m}^2\), corresponding to the rest of a larger school building. This was done both to maintain a realistic ratio between the smoke extracted from the fire room and the fresh air extracted from all other rooms, and to ensure a realistic ratio between the pressure build-up in the fire room and the total capacity of the AHU. As seen in Fig. 1, the ducts to and from the extra 300 \(\hbox {m}^2\) were located outside the mock-up building. The minimum, \(\dot{V}_{min}\), and maximum, \(\dot{V}_{max}\), design airflow rates depend on both room sizes and the intended usage and are shown in Table 2. All ventilation ducts were installed without insulation and suspended from the ceiling using standard ventilation brackets.
During the experimental campaign, two different types of DCV dampers were used. However, in the presented tests, all dampers exposed to smoke were of the brand name type LEO [13]. The airflow rate in this damper type is measured based on a thermal-mass flow sensor mounted on the outside of the damper. A small portion of the airflow is redirected to the sensor via measuring rods that form a cross upstream of the damper blade. Due to the thermal-mass flow sensor used, the measurement uncertainty can be expected to increase once the air temperature inside the dampers exceeds the normal operation range, which has an upper limit of \(50~^\circ {\hbox {C}}\). The LEO damper is manufactured in galvanized steel with an aluminium measurement unit. Other parts, such as the tubing leading from the rods to the thermal-mass flow sensor, nipples, and the motor casing are made of plastic, and the damper connection collars are fitted with EPDM rubber gaskets. The functionality of all dampers was checked after each test, and any damper that had been exposed to smoke was replaced with a brand-new one.
The ventilation system was equipped with all sensors needed for its operation: pressure and temperature sensors in the supply and extraction ducts and in the AHU, \(\hbox {CO}_2\) and temperature sensors in each room, and airflow sensors in each damper. The sensors connected to the BMS are not intended for use in fire tests, and the logging frequency was very low, in the order of half a minute. The uncertainty related to the BMS data is, therefore, expected to be high compared to the additional dedicated logging system that was installed. This logging system was used to log the gauge pressure in all three rooms, wall temperatures and the air temperatures at different heights and positions in each room, the temperature outside and inside of each damper, the carbon monoxide (CO), carbon dioxide (\(\hbox {CO}_2\)) and oxygen (\(\hbox {O}_2\)) concentrations in the supply and extraction valves, and the flow in each damper based on bidirectional probes. All temperatures were measured using 1.5 mm type-K thermocouples, with an uncertainty of \(\pm 3~^\circ {\hbox {C}}\) in the relevant range (15 - \(375~^\circ {\hbox {C}}\)). The inside temperature (Tinternal) of the dampers was measured by inserting a thermocouple into the centreline of the duct upstream of the dampers measuring rods. The outside temperature (Texternal) was measured by placing a thermocouple close to the control box. The position of these thermocouples can be seen by the example of the extraction damper to the classroom in Fig. 3. The thermocouples for the wall temperature measurement were stapled directly to the surface of the gypsum wall ensuring flush contact. The CO and \(\hbox {CO}_2\) concentrations were measured using a Dräger Polytron, and the \(\hbox {O}_2\) concentration was measured using a Dräger PIR7200. All pressure measurements with a Setra267 differential pressure gauge (0-500 Pa) have an uncertainty of ±0.1 %. A DustTrak DRX Aerosol Monitor Model 8533 was used to measure the mass concentration (\({\hbox {mg}}~{\hbox {m}}^{-3}\)) of particles with a size between 0.1 \({\upmu }{\hbox {m}}\) to 15 \({\upmu }{\hbox {m}}\) in the fire smoke. This measurement was made by continuously sampling air from the extraction air immediately before it went into the filter. To ensure that the particulate concentration would not exceed the instruments upper measuring limit, a in-house build dilution rig was used to add fresh air to the sample air stream. The amount of dilution with fresh air was adjusted with a flowmeter to achieve a 1:9 ratio between the two streams. Hence the measured particulate concentration was 1/10 of the real particulate concentration. The logging interval for the measurements was every second. Some of the measurements, such as the wall temperature and the thermocouple array measurements, are intended to make the results from the present experimental campaign accessible for future numerical work but do not directly contribute to the findings presented in this article.
Video recordings were made from two different angles, both in the classroom and the corridor, and from one angle in the office for each of the tests. The position of the different sensors and cameras can be seen in Fig. 2.
Fig. 2
Simplified sketch showing the instrumentation of the mock-up building and the ventilation system. The position of extraction and supply valves for all rooms are illustrated in yellow and blue, respectively. For the classroom, the supply and extraction ducts are also shown (Color figure online)
Thermocouple position for measuring the gas temperature inside (Tinternal) and outside, close to the control box, (Texternal) of the extraction DCV for the classroom
The present work presents two of the fourteen conducted tests: Test 10 and Test 14. A combination of propane and PU foam mattresses was used as fuel in both of these two tests. The mattresses that were used are the same type that is used for standardized fire testing water mist according to, among others, the test standard IMO 265 [14], having a size of 200 cm \(\times \) 90 cm \(\times \)10 cm and a mass of 5 kg. To increase the amount of smoke and reduce the heat release rate from the PU fire, the oxygen supply to this fire was reduced by building the mattress into non-combustible lightweight concrete blocks (Siporex). The mattress setup is shown to the left in Fig. 4. The same setup, with only small variations made in order to obtain preferable oxygen conditions for the fire, was used for all tests involving a mattress. To ignite the mattress, six rock wool (Rockfon) pieces in the size of 6 cm \(\times \) 6 cm \(\times \) 1.5 cm, placed inside a plastic bag filled with 120 mL heptane were used. The bag was placed on the floor next to the edge of the mattress. The ventilation system operated at \(V_{min}\) in the beginning of both tests. All doors in the mock-up building were kept closed during the tests, except when the personnel ignited the fires went out directly after ignition. A further description of the tests is provided in the following subsections.
Fig. 4
The fire sources used in the classroom for all tests with a single PU foam mattress. The mattress setup is to the left, and the propane burner is to the right
Test 10 was performed in the office using one encapsulated PU foam mattress and propane as fuel. The propane was burned in a 10 cm \(\times \) 10 cm \(\times \) 9 cm size burner completely filled with leca balls (light expanded clay aggregate). During a previous test, it was observed that burning the mattress next to the propane burner had a major impact on the gas flame by reducing its access to oxygen. To reduce the interaction between the two fire sources, a 60 cm high wall of Siporex blocks was set up between them. For Test 10, approximately 19 MJ were released from the mattress fire, and 96 MJ were released from the propane fire, based on the mass loss of the two fuel sources. The size of the propane fire was kept at a constant level of approximately 80 kW during the tests. Only the mass loss of the propane was measured continuously. The heat release rate (HRR) based on this measurement is shown in Fig. 5a.
The fire energy per base area in Test 10 was 10.5 \({\hbox {MJ}}~{\hbox {m}}^{-2}\), which is low compared to the fire energy that can be expected in a typical office. Eurocode 1 [15] states 420 \({\hbox {MJ}}~{\hbox {m}}^{-2}\) as fire energy per floor area for an office. Due to digitisation, it can be expected that a modern office contains less paper than the offices on which this value is based. Nevertheless, the available fire energy in a classroom or office is larger than the fires investigated in the tests. This was on purpose since the most frequent fires in Norwegian schools are small arson fires, often not spreading to other fuel than the object of origin [16].
The dampers of brand-type LEO for the supply and extraction of the office were located inside the office, i.e. inside of the fire room.
2.4.2 Test 14—Classroom Fire
Test 14 was performed in the classroom with two PU foam mattresses and propane used as fuel. The propane was burned in a 70 cm x 30 cm x 29 cm sized burner, filled up to 2/3 of its height with leca balls (light expanded clay aggregate). The propane burner used in the classroom can be seen to the right in Fig. 4.
For the first 10 minutes only the propane burner was used with an HRR of approximately 80 kW. After 10 minutes, the doors were opened, and the mattresses were ignited from both sides. At the same time, the HRR from the propane fire was increased to approximately 230 kW for the next 20 minutes before it was reduced again to 80 kW again. The reduction was done in order to sustain a stable flame for the gas burner after the propane fire showed signs of quenching. For Test 14, approximately 267 MJ were released from the mattress fire, and 539 MJ were released from the propane fire. The heat release based on the continuous propane flow measurement is shown in Fig. 5b.
The fire energy per base area was 14.4 \({\hbox {MJ}}~{\hbox {m}}^{-2}\), and hence larger compared to Test 10. However, it was still very small compared to the reference value of 285 \({\hbox {MJ}}~{\hbox {m}}^{-2}\) for a classroom [15].
The dampers of brand-type LEO for the supply and extraction of the classroom were located inside the corridor, i.e. outside of the fire room.
Fig. 5
HRR of the propane fire calculated based on the measured mass (Test 10) and the measured flow rate (Test 14)
This section presents the results and discusses the fire tests that were conducted. The uncovered failure mechanisms for typical DCV dampers are mainly discussed by the example of Test 10, which included all failure mechanisms observed during the entire experimental campaign. Some of the mechanisms were unique for Test 10, while others were observed in multiple other tests. Test 14, on the other hand, is used to evaluate the failure of the extraction filter and the performance of the entire ventilation system during the fire.
3.1 DCV Damper Failure Mechanisms
Figure 6 shows the gas temperature inside and outside of the supply and return damper during Test 10. The internal temperature in the extraction damper closely follows the external gas temperature since the dampers were located inside the office, i.e., the fire room. However, the internal gas temperature in the supply damper stayed low as long as the nominal flow rate was maintained (see Fig. 6b). Both dampers opened the damper blade position as intended once the fire was detected (i.e., at 0 seconds) to increase the flow rate from the initial \(\dot{V}_{min}\) to \(\dot{V}_{max}\). The BMS and the dedicated logging system for the fire tests reported this increase in flow rate. Note that there appears to be a delay between the two different flow rate measurements. This is caused by the very low logging frequency, 30 s and more between data points, of the BMS and two different clocks, which made synchronising these two logging systems challenging.
Fig. 6
Volume flow rate as reported by the BMS (\(\dot{V}_{BMS}\)) and calculated based on measurements with a bidirectional probe (\(\dot{V}_{effective}\)), damper blade position (100 % is fully open and 0 % fully closed) and gas temperature inside (\(\hbox {T}_{internal}\)) and outside (\(\hbox {T}_{external}\)) the damper. The time axis is shifted so the fire detection by the BMS corresponds to 0 seconds
The flow rate measurements in the extraction and supply damper deviate from each other after about six and seven minutes, respectively. At this point, it can also be noted that the extraction damper opened to 100 %, while the supply damper closed completely. The deviating flow measurements indicate a failure of the measurement units of the DCV dampers. It is not clear why the dedicated logging system still measured flow in the supply line after the damper was reported to be closed. One explanation could be a faulty damper location signal. However, the damper was inspected after the test was finished and appeared closed. The measurement could also indicate that a significant leak flow was present, caused by the fault in the extraction damper, which led to the full opening of the damper and, hence, the creation of an underpressure in the room. This theory is supported by the room pressure measurement (see Fig. 7), which falls to zero once the dampers fail. The logging system was not set up to record negative pressures. This leak flow of almost 60\({\hbox {m}}^{3}~{\hbox {h}}^{-1}\) is also supported by the propane fire burning for around 15 minutes after the supply damper was closed. The propane fire required an estimated 77\({\hbox {m}}^{3}~{\hbox {h}}^{-1}\) air supply.
Fig. 7
Gauge pressure in the classroom, the office and the corridor during Test 10. The time axis is shifted so the fire detection by the BMS corresponds to 0 seconds
The visual inspection of the dampers showed that the tubing that leads air from the inside of the dampers to the measuring unit located on the outside of the damper had melted. The melted tubing of the supply damper is shown in Fig. 8b. Figure 8a shows a new damper for comparison before the start of Test 10. The inspection showed furthermore that the cabling to the dampers control unit was damaged (see Fig. 8c). Hence, the results from Test 10 alone are not conclusive as to which damage caused the complete closing and opening of the dampers, respectively. However, four other tests during which the cables to the control unit were not damaged, including Test 14, showed that damaging the tubing or its connections due to heat exposure causes the extraction dampers to open fully. The dampers in these tests were located outside the fire room, which helped reduce the heat exposure to the cabling and the measuring unit. However, the plastic tubing of the extraction dampers was still exposed to enough heat to get damaged, leading to no flow through the measuring unit. The DCV damper consequently tried to increase the flow by opening fully. However, the supply damper’s closing was only observed in Test 10. Another failure mechanism that was only observed in Test 10 was the short-circuiting in the 24 V power supply, which caused the BMS to lose contact with all dampers in the building. Therefore, only the last obtained BMS data is shown in Fig. 6 after 10 min.
Fig. 8
Pictures of the new extraction damper in the office before the start of Test 10 (a), the melted tubing of the supply damper to the office after Test 10 (b) and damaged cabling and measuring unit of the supply damper to the office after Test 10 (c)
The effect of the extraction damper’s opening and the supply damper’s closing can also be seen in the gas measurements that were taken in the supply and extraction air and are most notable in the CO concentration (see Fig. 9). It can be seen that the CO concentration slightly drops after it had been increasing up until six minutes, at which point the effective extraction flow increased due to the extraction damper being opened fully. Two minutes later, at eight minutes, the supply damper had closed completely, causing the effective extraction flow rate to decrease again and reducing the oxygen supply to the office. This is reflected in the increasing CO concentration measured in the extraction air. The concentration of CO, \(\hbox {O}_2\), and \(\hbox {CO}_2\) in the supply duct did not change during the test.
The gas measurements in the supply air to the office, i.e., the fire room during Test 10, showed a constant \(\hbox {O}_2\) concentration and no \(\hbox {CO}_2\) or CO. Hence, the pressure build-up in the fire room was not sufficient to cause smoke to spread via the supply ventilation duct. However, the pressure build/up was sufficient to cause leakage from the office to the adjacent corridor. This was confirmed by the \(\hbox {CO}_2\) room sensor in the corridor, which reached its maximum measuring concentration of 2 000 ppm (not shown here). A very slight reduction in visibility was also observable in the video recordings of the corridor. The reduction was so little that it was mainly noticeable around the installed light sources.
It is important to discuss the fact that the identified failure mechanisms, which led to the unintended closing and opening of DCV dampers, increase the risk of smoke spreading through the supply duct, even though no smoke spreading through the ventilation ducts was observed in Test 10. Since both supply and extraction dampers use the same damper design, a scenario in which the supply damper opens fully and the extraction damper closes fully cannot be ruled out. This would be the worst-case scenario in practice, as opening the supply damper would reduce the duct pressure while closing the extraction damper would increase the room pressure.
Fig. 9
\(\hbox {O}_2\) (cyan), \(\hbox {CO}_2\) (purple) and CO (pink) concentration measured in the supply (dashed line) and extraction (solid line) duct during Test 10. The time axis is shifted so the fire detection by the BMS corresponds to 0 seconds
Test 10 led to the lowest soot concentration in the extraction air, with an average PM2.5 particle (< 2.5 \({\upmu }{\hbox {m}}\)) concentration of 0.48 \({\hbox {mg}}~{\hbox {m}}^{-3}\). No clogging of the measuring rods or tubing of the DCV dampers was observed during the test. This was also the case for all other tests, including Test 14, which resulted in the highest average PM2.5 concentration of 18.8 \({\hbox {mg}}~{\hbox {m}}^{-3}\). The different in soot production between these two tests is also reflected in the mass increase of the extraction filters that were measured during Test 10 and Test 14, with 11 g for the first and 474 g for the latter.
The fact that none of the tests led to the clogging of the damper measuring unit (including rods and tubing) before other parts of the ventilation systems failed, indicates that heat exposure is a larger threat than soot exposure. All dampers were brand new when they were installed in the mock-up building. To what extent dust accumulation over several years may affect the risk of clogging the measuring unit with soot was therefore not assessed.
3.2 Effect of Failure on Smoke and Pressure Control
This section discusses how the failure of an individual damper, which has been identified as a possible failure mechanism in the previous section, can affect the performance of the entire DCV system and, hence, impact the pressure and smoke control during a fire. This is done using Test 14 as an example. The later stage of Test 14 also provides insight into how gradual clogging of the extraction filter affects the system’s performance.
Figure 10 shows the supply and extraction volume flows in the classroom, the office and the corridor. The flow rates in the classroom (Fig. 10a), the fire room in Test 14, are based on the bidirectional probe measurements since the BMS data cannot be used once the DCV dampers get damaged by the fire exposure. The volume flow rates in the other rooms (Fig. 10b and c) are reported by the BMS since the dampers in these rooms are not exposed to the fire and therefore provide reliable data. This is not true for Test 10 during which the contact with all dampers was lost due to short-circuiting, as discussed in the previous section. It can be seen that the ventilation system increases the flow rates in the classroom and the office to the respective \(\dot{V}_{max}\) values after the fire is detected. It can be seen that also the DCV dampers in the corridor, which has a constant nominal volume flow rate (i.e., \(\dot{V}_{max} = \dot{V}_{min}\)), reacted by changing the damper blade position in order to account to the changed pressure conditions in the ventilation system. The ventilation system maintained balanced ventilation in all rooms, including the rest of the school, throughout the first ten minutes, i.e. the duration of the pure propane fire.
Fig. 10
Supply and extraction volume flow rate and damper opening degree during Test 14 for the classroom,i.e., the fire room, (a), the office (b) and the corridor (c). The flow rates for the classroom are based on the bidirectional probe measurements, while the flow rates for the office and classroom are reported by the BMS. The time axis is shifted so the fire detection by the BMS corresponds to 0 seconds
Figure 11 shows a moderate pressure increase up to between 25 and 50 Pa during the initial phase of the fire. The pressure slowly decreased once the heat release rate of the propane burner was kept constant, as leakage to the other rooms and the exterior caused the pressure difference to equilibrate. After around ten minutes, one external door to the corridor and the classroom door were opened to ignite the PU foam. The opening of the doors causes a short pressure peak in the corridor measurement and, consequently, a pressure drop in the classroom measurement. The pressure in the classroom rapidly increased, with the highest peak reaching 197 Pa, once both mattresses were ignited, and the propane flow rate was increased to approximately 0.298 \({\hbox {kg}}~{\hbox {min}}^{-1}\) (±6 %). It can be seen in Fig. 10 that all dampers are reacting to the changed pressure conditions in the building in order to maintain balanced ventilation at nominal volume flow rates. Ten minutes after the fire size was increased, at around 20 minutes, it can be seen that the extraction DCV damper from the classroom opens fully (see Fig. 10a). This was caused by the same failure mechanism as already observed in Test 10, the failure of the tubing and, consequently, the wrong flow measurement reported by the extraction damper. The classroom extraction damper is located upstream of the office extraction damper. The sudden pressure change upstream of the office caused the extraction flow rate from the office to drop, which the ventilation system tried to compensate for by opening the extraction damper fully and increasing the extraction fan speed to 100 %. The extraction fan operated at around 94 % during the first ten minutes. However, the system was not able to maintain the required extraction flow rate from the office, which dropped to around 150 \({\hbox {m}}^{3}~{\hbox {h}}^{-1}\), 60 % of the nominal flow rate. Hence, the fire in the classroom caused unbalanced ventilation in a room which is not directly exposed to the fire (see Fig. 11), which is an important finding of the present work since the pressure buildup potentially can impair the ability to evacuate the building [3] if the pressure difference is sufficiently large. How large the pressure difference in a specific building would be depends, among others, on the room dimensions, leakage area and design of the ventilation system.
It can be observed that the extraction dampers in the corridor (Fig. 10c) and the rest of the school gradually opened once the classroom damper failed. Unlike in the office, opening these other dampers was sufficient to maintain \(\dot{V}_{max}\). The extraction damper for the rest of the school reached its fully open position at around 28 minutes. At this point, it is no longer possible to assess the pressure build-up in the building quantitatively. The air exchange with the rest of the school, i.e., the ambient, cannot introduce any pressure build-up. The pressure imbalance is expected to be amplified in a real building, where all air exchange is with enclosed rooms. The pressure decrease in all rooms at around 32 minutes corresponds to the reduction of the propane flow rate as described in Sect. 2.4.
Fig. 11
Gauge pressure in the classroom, the office and the corridor during Test 14. The time axis is shifted so the fire detection by the BMS corresponds to 0 seconds
While the failure of the extraction damper from the classroom had a relatively instant effect on the ventilation system, the failure of another component, the extraction filter, had a more gradual impact on its performance. Figure 12 shows the supply and extraction volume flow rate in the AHU, the pressure loss across the extraction filter and the gas temperature upstream of the filter during Test 14. Before the first fire test, it was noted that the supply flow rate measured in the AHU was larger than the total flow to the building, i.e., the sum of all DCV dampers. The flow deviation was reduced by sealing some leaks that had been identified but could not be eliminated completely. Hence, a minor difference between the supply and extraction flow rate can be seen in Fig. 12 even though the ventilation is balanced initially. Following the initial rise caused by the increasing volume flow rates, the pressure drop across the extraction filter continued to climb gradually. The slope of the pressure loss curve increased significantly after the ignition of the PU mattresses and the rise in propane flow rate, reaching the measurement limit of 250 Pa after approximately 35 minutes. After that, the effect of the filter clogging can only be observed indirectly through the decreasing extraction volume flow rate reported by the AHU and the correlated opening of the individual dampers. At 45 min, it can be seen that the rate with which the extraction damper in the corridor (Fig. 10c) opens increases. The pressure in all rooms increases at the same time, and pressure fluctuations can be observed (Fig. 11). The extraction damper in the corridor is fully open after about 57 minutes. This means that all extraction dampers are fully open at this point, and the extraction fan operates at its full capacity, causing imbalanced ventilation in all rooms.
Fig. 12
Supply and extraction volume flow rate in the AHU, pressure loss across the extraction filter and gas temperature upstream of the filter during Test 14. The time axis is shifted so the fire detection by the BMS corresponds to 0 s
The flow measurement in the AHU is based on differential pressure measurements. Hence, the uncertainty attributed to the extraction flow rate increases with the gradual clogging of the extraction filter. Due to the filter clogging, it is possible that the pressure difference generated by the extraction fan is not or is only partially converted to an actual flow through the extraction duct. The lowest volume flow rate reported by the AHU was around 2 800 \({\hbox {m}}^{3}~{\hbox {h}}^{-1}\), 62 % of the nominal total extraction flow rate, highlighting the impact of soot exposure and the importance of bypass systems that lead the smoke around the filter during the event of a fire. On the other hand, the extraction gas temperature was below \(70~^\circ {\hbox {C}}\), its upper limit for normal operation, throughout the entire fire test. Heat exposure depends on many factors, such as the size of the fire, the fraction of smoke in the extraction air, the design of the ventilation system, and the fire location.
The \(\hbox {CO}_2\) concentration in the corridor reached the measurements limit of 2 000 ppm after 20 minutes, and the visibility was clearly reduced, but it was still possible to see the outline of the door on the opposite side of the corridor, i.e., in a distance of 15 m, in the video footage. This observation is, however, only indicative since the light conditions in the corridor were not controlled, and the smoke density was not measured. Furthermore, it cannot be quantified how much smoke spread to the corridor during the door opening after ten minutes.
Test 10 and Test 14 led both to smoke leaking into the adjacent room, the corridor. This was also the case for all other tests in the test campaign that had a fire load of 100 MJ or more. The video recording from one of the tests showed that the lower part of the office door was an area where smoke leaked into the corridor. Smoke spreading through the supply or extraction ducts was not observed in any of the tests. However, Test 10 showed that heat exposure caused by a fire can lead to a DCV damper closing. Since the same damper type is used for supply and extraction, it cannot be ruled out that an extraction damper closes down, leading to a pressure increase in the fire room comparable to the pressure build-up when employing the compartmentation strategy. Brohez and Caracita [4] measured, for example, peak pressures between 420 and 750 Pa for tests with active ventilation (without DCV dampers) and pressure peaks between 870 and 2 035 Pa for tests with compartmentation. Hostikka et al. [3] found that changing from active ventilation to compartmentation increased the peak pressure from 46 - 2 316 Pa to 72 - 7 069 Pa. However, in the case of compartmentation, both the supply and extraction dampers are closed by fire dampers. The supply damper is supposed to be open for the extraction strategy. Hence, the ventilation system needs to be able to provide sufficient pressure to prevent smoke from spreading via the supply duct, which is more challenging for high pressures in the fire room. The risk of smoke spreading through the supply duct is, therefore, increased if a failure leads to the closing of an extraction damper. Smoke spreading through the extraction duct also has to be considered, based on the finding that the gradual clogging of the extraction filter can significantly reduce the extraction volume flow. However, the problem of filter clogging can be circumvented by incorporating a bypass into the ventilation system.
4 Conclusions
Two full-scale fire tests, out of a test campaign with 14 tests, in a mock-up school building with multiple rooms, a classroom, an office and a corridor were presented. The building was equipped with a fully functional damper-optimised DCV system with a capacity to serve a total of 450 \({\hbox {m}}^{2}\) to provide a realistic basis for the fire tests. The tests were designed to trigger potential faults of the DCV system and to observe their impact on the ventilation system performance during the fire. The main conclusions from the fire tests are:
Different plastic components of non-fire-rated DCV dampers meld, as expected when exposed to sufficient heat from a fire. The most frequently observed failure mechanism was the melting of the tubing and connection nipples. However, the novel conclusion is that this can lead to an unintended full closing or opening of the damper, caused by faulty flow measurements. This observation can have severe practical implications. Closing the extraction damper due to a failure would increase the room pressure and, therefore, the risk of smoke spreading via the supply duct. The risk of smoke spreading is also increased if the fire exposure causes the supply damper to open, as this would decrease the duct pressure. The worst-case scenario is a closing of the extraction damper combined with an opening of the supply damper.
The failure of a single damper in a damper-optimised ventilation system can affect the ventilation balance and, hence, the pressure control in multiple rooms that are connected to the same ventilation system. The important observation is that this can also affect rooms which are not directly adjacent to the fire room and, hence, impair the evacuation of the building. The extent of the pressure imbalance depends on multiple factors, such as the building and ventilation system design and the fire development.
Some failure mechanisms affect all dampers at the same time. The exact mechanism behind such errors is unknown since several errors can occur simultaneously, but short-circuiting was observed in the 24 V power supply.
The extraction dampers are exposed to a greater heat and smoke load than the supply dampers, which are cooled from the inside as long as the supply flow continues. This means that supply dampers can, in practice, be protected by placing them outside of the room to which they are supplying air to (e.g. in the corridor). However, extraction damper designs need to be modified in order to be able to document their performance during a fire.
A gradual clogging of the extraction filter was observed in the fire tests. How quickly the filter clogging affects the ventilation system depends on many factors, such as the ratio between the fire size and the air handling unit’s capacity, fuel composition, filter area and filter condition. Since the extraction airflow was greatly reduced in several tests, up to around 40% of the maximum projected air quantity, it is concluded that gradually clogging the extraction filter will increase the risk of smoke spreading via the extraction duct.
How long the ventilation system can provide sufficient high supply pressure to prevent smoke from spreading through the supply duct and to compensate for an increasing pressure drop over the extraction filter depends, among others, on the remaining capacity of the AHU. The remaining capacity of the AHU is larger when the ventilation system operates at \(\dot{V}_{min}\). However, the extraction of smoke that has leaked to adjacent rooms will also be reduced. Future work should, therefore, investigate if different ventilation strategies, i.e., running the ventilation system at \(\dot{V}_{min}\) or other flow rates below \(\dot{V}_{max}\), could reduce the clogging rate of the filter and affect the conditions in the fire room in a beneficial way. If complex damper-optimized ventilation systems are intended to be used for smoke extraction during a fire it is necessary to investigate design changes that make it possible to document the fire resistance of the ventilation system. This can include material changes to high-temperature resistance materials or mechanisms that would fix the damper position of the DCV damper during a fire (i.e., acting as VAV dampers), such that faults in the power supply, measuring unit or communication connections would not lead to unintended changes in the damper position. In order to avoid clogging of the extraction filter a bypass needs to be installed.
Acknowledgements
This research has been financed by the project "BRAVENT - Efficient smoke ventilation of small fires" funded by the Research Council of Norway, grant no. 321099 and its project partners. The experimental campaign has been conducted in collaboration with the project "Fire evacuation during lithium-ion battery fires in electric scooters" funded by the Norwegian Directorate for Civil Protection and the Norwegian Building Authority. The authors, furthermore, thank Interfil and Autronica Fire and Security for donating the filters and the fire alarm system respectively to the project.
Declarations
Competing interest
The authors have no financial or non-financial competing interests in this work.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Mysen M, Schild PG, Cablé A (2014) Demand-controlled ventilation—requirements and commissioning. Tech Rep. ISBN 978-82-536-1414-4
2.
Gordonova P (1997) Spread of smoke and fire gases via the ventilation system. Tech Rep. TABK–97 / 1011. Lund University
3.
Hostikka S, Janardhan RK, Riaz U, Sikanen T (2017) Fire-induced pressure and smoke spreading in mechanically ventilated buildings with air-tight envelopes. Fire Saf J 91:380–388CrossRef
4.
Brohez S, Caravita I (2020) Fire induced pressure in airthigh houses: experiments and FDS validation. Fire Saf J 114:103008CrossRef
5.
Li J, Beji T, Brohez S, Merci B (2020) CFD study of fire-induced pressure variation in a mechanically-ventilated air-tight compartment. Fire Saf J 115:103012CrossRef
6.
Prétrel H, Le Saux W, Audouin L (2012) Pressure variations induced by a pool fire in a well-confined and force-ventilated compartment. Fire Saf J 52:11–24CrossRef
7.
Li J, Prétrel H, Suard S, Beji T, Merci B (2021) Experimental study on the effect of mechanical ventilation conditions and fire dynamics on the pressure evolution in an air-tight compartment. Fire Saf J 125:103426CrossRef
Fjærestad JS, Yang A, Meraner C, Dovran F, Olsen J (2024) BRAVENT - Storskala branntester (del 1): Brannytelse for ikke-brannklassifiserte ventilasjonskomponenter. Tech Rep. RISE-rapport 2024:37. RISE Fire Research
10.
Meraner C, Fjærestad JS (2024) BRAVENT—Storskala branntester (del 2): Effekt av ventilasjonsstrategi på røykspredning og trykkontroll i en mock-up skolebygning. Tech Rep. RISE-rapport 2024:38. RISE Fire Research
11.
Meraner C, Floyd J, Chaudhari DM, Beji T, Fjærestad JS (2024) Modelling a damper-optimized demand control ventilation system during a fire. To be submitted
12.
ISO 9972:2015 Thermal performance of buildings—determination of air permeability of buildings—fan pressurization method (2015)
IMO Res. MSC.265(84) Annex 14 Amendments to the revised guidelines for approval of sprinkler systems equivalent to that referred to in SOLAS regulation II-2/12 (Resolution A.800(19)) (2008)
15.
EN 1991 1-2 (2002) Eurocode 1: Actions on structures—Part 1-2: general actions—actions on structures exposed to fire. Tech Rep. European Committee for Standardisation (CEN), Brussels, Belgium (2002)
16.
Yang A et al (2022) Fire safety of ventilation systems and fire incidence reports in Norwegian schools. E3S Web Conf 356:02001CrossRef
