- Email: [email protected]
Required response time for variable air volume fume hood controllers
PII: S0003-4878(99)00082-4
Ann. occup. Hyg., Vol. 44, No. 2, pp. 143±150, 2000 # 2000 British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain. 0003±4878/00/$20.00
Required Response Time for Variable Air Volume Fume Hood Controllers LARS E. EKBERG* and JAN MELIN Department of Building Services Engineering, Chalmers University of Technology, S-412 96 GoÈteborg, Sweden This paper describes results from tests made with the aim of investigating how quickly the exhaust air ¯ow rate through fume hoods needs to be controlled in order to prevent contaminants from leaking out of the fume hood and putting the safety of the laboratory personnel at risk. The measurements were made on a laboratory fume hood in a chemical laboratory. There were no other fume hoods in the laboratory, and the measurements were made without interference from persons entering or leaving the laboratory or walking about in it. A tracer gas method was used with the concentration of dinitrogen oxide (N2O) being recorded by a Foxboro Miran 101 infra-red gas analyser. In parallel with the tracer gas measurements, the air velocity through the face opening was also measured, as was the control signal to the damper controlling the air ¯ow rate. The measurements show an increased outward leakage of tracer gas from the fume hood if the air ¯ow rate is not re-established within 1±2 s after the sash is opened. If the delay exceeds 3 s the safety function is temporarily defeated. The measurements were made under virtually ideal conditions. Under more typical conditions, the fume hood could be exposed to various other external perturbations, which means that the control system should re-establish the correct exhaust ¯ow more quickly than indicated by the measurement results obtained under these almost ideal conditions. # 2000 British Occupational Hygiene Society. Published by Elsevier Science Ltd. All rights reserved. Keywords: laboratory ventilation; control system; face velocity; fume hoods
INTRODUCTION
Historically, laboratory fume hoods have usually operated with a constant air ¯ow rate. However, it has become increasingly common in present-day modern laboratories to control the exhaust air ¯ow rate from fume hoods in order to maintain a constant air velocity through the face, regardless of the position of the sash. There are two main methods that are normally used for this purpose: one measures the air velocity, while the other responds to the position of the sash, with the correct ¯ow rate being set from one of these parameters. When using the face velocity as a control parameter, there is generally some form of velocity sensor ®tted to the inside of the fume hood. However, there are a number of factors that can aect the Received 8 December 1998; in ®nal form 12 July 1999. *Author to whom correspondence should be addressed. Tel.: +46-31-772-1162; fax: +46-31-772-1152; E-mail: [email protected] 143
measured velocity so that it is not representative of the average velocity through the face opening. Examples of such factors include equipment inside the fume hood upsetting the ¯ow pattern around the sensor, the creation of draughts within the hood due to the movements of the operator, or a draught across the face opening. One further important condition to consider when measuring the air velocity close to the fume hood wall is that the air velocity through the face opening is not necessarily always uniformly distributed. Results from air velocity measurements, carried out using a hot wire anemometer at several points for three dierent face openings in a fume hood installed in a laboratory, reveal that there may be positions where the measurement does not correspond fully with the average face velocity. This means that there may be a need to correct the measured air velocity in order to get a value representative of the average face velocity. This eect can be quite pronounced in measurement positions close to the edge of the sash and along the sides
144
L. Ekberg and J. Melin
and the lower edge of the fume hood opening, where also substantial velocity ¯uctuations due to turbulence can be expected. Using the position of the sash as the control parameter employs the relationship that the area of the face opening is proportional to the position of the sash. In purely practical terms, this is usually sensed by securing a wire to the sash to drive a potentiometer, the signal from which then controls a damper or a fan to provide the correct ¯ow rate. An important dierence between the two control principles is their speed of operation. In papers by Rabiah et al. (1989) and Lacey (1989), results are presented from trials where the response times for dierent types of commercially available control systems were tested. The results show that systems that employ the sash position as the control parameter generally are faster in their response than systems that use the air velocity as the control parameter. There is no doubt that, in almost all cases, it is an advantage to control the exhaust air ¯ow from the fume hood. Similarly, there is no doubt that dierent types of control systems react at dierent speeds. How quickly does the ¯ow need to be controlled to ensure that the fume hood provides adequate protection? A study by Ahmed and Bradley (1995) describes measurements intended to investigate the necessary response time for fume hood control systems. The measurements were based on ANSI/ASHRAE Standard 110-1995, using sulphur hexa¯uoride as the tracer gas and dosing it into the fume hood via an ASHRAE Standard tracer gas ejector, with the measurement point positioned at the breathing zone of a dummy. The experiments were carried out under ideal conditions in two laboratories on fume hoods without existing ventilation control. Instead, ¯ow control was eected by manual control of the fan motor speed. A portable electron capture cell detector was used for analysis of the gas concentration, recording the results in one laboratory on a strip chart recorder and in the other on a PC. Opening the sash from 25% to fully open, without increasing the ventilation ¯ow rate, reduced the air velocity through the face opening from 100 fpm (0.5 m sÿ1) to 25 fpm (0.13 m sÿ1). The time from the sash reaching fully open until tracer gas was recorded varied over a wide interval. It was shown that no gas leaked out before about 2±3 s from the sash being fully opened. In another series of measurements, the sash was opened from 25% to fully open and, after a certain time, the speed of the fan was increased to restore the average front velocity of 100 fpm. In this case, the time between the sash being fully opened and the instance when the speed of the fan needs to be
increased in order to avoid outward leakage was of the order of 2±3 s. The present paper describes results from another investigation made with the aim of investigating how quickly the ventilation air ¯ow rate needs to be controlled in order to prevent contaminants from leaking out of the fume hood and putting the safety of the laboratory personnel at risk. EXPERIMENTAL PROCEDURE
The measurements were made on a laboratory fume hood of the type shown in Fig. 1. There were no other fume hoods in the laboratory, and the measurements were made without interference from persons entering or leaving the laboratory or walking about in it. No equipment was placed in the fume hood. The fume hood had an aperture width of 102.5 cm and a fully open face height of 74 cm. The sash was intentionally designed so that it could not be completely closed: when pulled down as far as possible, a gap was left at the bottom in order to maintain a basic through ¯ow rate of 100 m3 hÿ1. The fume hood had a control system for maintaining a constant average face velocity of 0.5 m sÿ1, regardless of the position of the sash. The control system was of the sash position sensing type, with the control signal from the potentiometer that sensed the sash position controlling a valve for adjustment of the air ¯ow rate. Moving the sash to
Fig. 1. Sketch of the aerodynamic type of laboratory fume hood studied.
Variable air volume fume hood controllers
its fully open position resulted in an increase in the air ¯ow rate towards 1365 m3 hÿ1. The tracer gas method was used in order to determine how quickly the exhaust air ¯ow rate needed to be regulated. Nitrous oxide, N2O, was used as the tracer gas, with the concentration being recorded by a Foxboro Miran 101 IR gas analyser. According to the manufacturer, the detection limit for N20 using this instrument is 0.5 ppm. The measurements were made with a person standing in front of the fume hood, making no movements apart from opening the sash. The measurement probe was placed in two positions: in the wake region in front of the person standing by the fume hood, 3 cm above the airfoil and 1 cm from the plane of the sash, and in the person's breathing zone, 160 cm above the ¯oor and 5 cm from the plane of the sash. The time taken until the tracer gas was registered by the gas analyser was corrected to allow for the instrument's dead time, that is the time it took for the gas to travel from the measurement probe to the instrument. The dead time was found by using the Miran 101 gas analyser together with another instrument that has a negligible inertia (Melin, 1997). Simultaneous measurements with these two instruments in a duct into which tracer gas was dosed revealed that the dead time was 1.5 s. The tracer gas was dosed through a linear spreader and, in certain cases, through a point spreader. Both the point spreader and the linear spreader were used for the measurements in the wake region of the person standing in front of the fume hood. The point spreader was positioned 15 cm into the fume hood, at a height of 25 cm above the working surface. The line spreader was also placed 15 cm into the fume hood, but at a height of 10 cm above the working surface. For the measurements in the breathing zone, only the linear spreader was used, being positioned 15 cm into the fume hood and at a height of 37.5 cm above the working surface. The tracer gas ¯ow rate through the point spreader was 87 l. hÿ1, while that through the linear spreader was 120 l. hÿ1. The gas ¯ow was measured using a rotameter. In parallel with the tracer gas measurements, the air velocity through the face opening was also measured, as was the control signal to the control valve. The face velocity was measured using an air velocity sensor incorporating a thermistor, positioned 4 cm below the edge of the sash. With the time constant used for the tests, the response time of this instrument was 0.2 s. A switch was ®tted to the sash in order to be able to measure exactly when it opened. The signals from the gas analyser, the air velocity sensor, the control device and the switch were scanned at a rate of 10 values sÿ1 and stored in a computer. The ®rst stage of the measurements involved investigating the speed of response of the existing
145
control system. In order to estimate the time for the control valve and exhaust air ¯ow rate to change to the required level after a change in the position of the sash, measurements were made that involved quickly opening the sash while logging the control signal and the air velocity through the sash opening. These measurements were made from the closed sash position to three opening heights: onethird open (1/3), two-thirds open (2/3) and fully open (1/1). The sash was opened at a rate of about 0.5 s from the closed to the fully open position. The next part of the investigation was designed for determination of how quickly the air ¯ow through the fume hood needed to be controlled in order to prevent tracer gas from leaking out. For these measurements, the tracer gas was dosed inside the fume hood, and the sash was opened from the closed position to the fully open position at a rate equivalent to 2 s. The reason for taking such a long time to open the sash was to avoid the actual act of opening the sash from creating turbulence in the face opening and thus causing an outward leakage of tracer gas. During the ®rst series of measurements, the control system was completely disconnected, that is there was only the basic ventilation ¯ow through the fume hood. The purpose of these measurements was to determine the time it took for the tracer gas to leak out from the fume hood after the sash was opened. Repeated measurements were made, using both the point spreader and the linear spreader, and with the measurement probe in the wake region of the person standing at the fume hood. In order to prevent a build-up of tracer gas in the laboratory when making the measurements, the control system was activated as soon as the gas analyser had detected the gas. The existing control system was in operation for the second series of measurements. These measurements were intended to determine how quickly the system needed to react and set the correct air ¯ow rate in order to avoid outward leakage from the fume hood. During the ®rst set of measurements in this series, the control system was allowed to respond in its normal manner. The sash was repeatedly opened and the tracer gas concentration at the measurement point was recorded. During the next set of measurements, the control system was deactivated for 1 s, that is a delay of 1 s was introduced before the system was allowed to respond and adjust the air ¯ow in the normal way. Again, the sash was opened repeatedly and the tracer gas concentration recorded in the same way as before. The same overall procedure was then followed, but with respective delays of 2, 3, 4, 5 and 6 s. The described procedures were carried out using both point spreaders and linear spreaders for the measurement points in the wake region of the operator and in the breathing zone.
146
L. Ekberg and J. Melin Table 1. Delay times for the existing control system
Sash opening
1/3 2/3 1/1
Sash position command
Face velocity
Delay (s)
Standard deviation (N = 15)
Delay (s)
Standard deviation (N = 15)
0.9 1.4 1.8
0.07 0.17 0.07
1.7 1.7 1.9
0.31 0.15 0.16
RESULTS AND DISCUSSION
Existing control system Table 1 shows the delay times that could be determined from the measurements carried out with the existing control system in operation. The delay time for the sash position command is de®ned as the time from when the sash was opened until the control signal had reached a stable level. The delay time for the face velocity is de®ned as the time from when the sash was opened until the air velocity at the measurement point had ®rst reached 0.5 m sÿ1. The table shows that, for the worst case of when the sash was opened to its fully-open position, the air velocity through the face opening responded in less than 2 s from starting to open the sash. Operation without air¯ow control Figure 2 is an example of a test, showing how the air velocity drops and then stabilises at a low velocity when there is no increase in the exhaust air ¯ow rate due to the control system having been disengaged. A total of 10 measurements of the same type were carried out each with a linear spreader and a point spreader in use. The results are discussed in the following. The time taken for the tracer gas to reach the instrument varied somewhat, depending on
whether the gas was dosed through the point spreader or through the linear spreader. When dosing the gas through the point spreader, it was found that the average value of the time between the sash being fully opened and the gas reaching the instrument was 2.6 s, with a standard deviation of 0.9 (N = 10). The shortest time was 1.2 s and the longest time was 4.2 s. When dosing the gas through the linear spreader, it was found that the average value of the time, as de®ned above, was 3.0 s, with a standard deviation of 0.9 (N = 10). In this case, the shortest time was 1.7 s and the longest time was 4.6 s. The measured results thus provide an average value for all the measurements (that is with both point spreader and linear spreader) of the time from the sash being fully opened until the tracer gas had reached the instrument, of 2.8 s, with a standard deviation of 0.9. The times noted above are de®ned as being the time from when the sash reached its fully open position until the tracer gas reached the measurement point. Alternatively, it could be de®ned as the time between starting to open the sash and the gas reaching the measurement point. By this de®nition, the average value for both the point spreader and the linear spreader would be 4.8 s, with a standard deviation of 0.9. The fact that an outward leakage of the tracer
Fig. 2. Air velocity in response to opening the sash with the control system out of use.
Variable air volume fume hood controllers
gas occurs is naturally due to the eect of the average velocity through the face opening falling to such a low value that it is insucient to prevent undesired spread of contaminants. Ljungqvist (1991) showed that reducing the air velocity through the face opening by 20±40% of the nominal value of 0.5 m sÿ1 results in an increased outward leakage. Figure 2 shows how the air velocity through the face opening falls when the sash is opened without increasing the air ¯ow rate. After only 0.5±1.0 s from starting to open the sash, the velocity has dropped by about a factor of two, and consequently, there is a risk of increased outward leakage. The results from measurements with the control system completely out of operation and measuring the time until tracer gas is detected, indicate how quickly a control system needs to respond and start to adjust the air ¯ow rate in order to avoid outward leakage of contaminants. However, further tracer gas measurements carried out at various speeds of the air ¯ow rate control are needed in order to obtain more detailed information about what demands should be made on the control system. Operation with dierent time delays of the control signal This section gives the results from measurements carried out with the existing control system in operation. These measurements were intended to determine how quickly the system needed to react and set the correct air ¯ow rate in order to minimise the outward leakage from the fume hood. During the ®rst set of measurements in this series, the control system was allowed to respond in its normal manner. The sash was repeatedly opened and the tracer
147
gas concentration at the measurement point was recorded. As mentioned previously, the sash was moved from the closed to the fully open position in 2 s. Figure 3 shows the response for the case with no time delay before the control system was allowed to start operating. During the next set of measurements, the control system was deactivated for 1, 2, 3, 4, 5 and 6 s. These measurements were performed in sequences taking about a minute. This meant that, if and when leakage occurred, any tracer gas that had leaked out was evacuated from the laboratory before the next sequence started, thus preventing any build-up of concentration of tracer gas in the laboratory. Figure 4 shows an example of the sequence of changes in the sash position. The ®gure shows the response with a 4-s delay before the control system was activated. Furthermore, it can be shown that the delay of the control signal is approximately equal to the delay between the sash being fully opened and the air velocity being reestablished at its nominal value of 0.5 m sÿ1. Figure 4 shows that, with a 4-s delay in the control system, the tracer gas concentration in the wake region of a person standing in front of the fume hood may reach a few hundred ppm. For comparison it should be noted that the concentration was less than one hundred ppm in the return air from the fume hood when the sash was fully open. Moreover, the measurements in the breathing zone also show a considerable outward leakage of tracer gas with a control system delay of 6 s, that is if there is a delay of 6 s from the sash reaching the fully open position (or 8 s from the time of starting to move the sash) before the correct air velocity is re-established, there will be a considerable outward leakage.
Fig. 3. Control signal and air velocity when opening the sash from closed to fully open over a period of 2 s, with no delay in the control system.
148
L. Ekberg and J. Melin
Fig. 4. An example of a series of measurements of a 4-s delay of the control system and with the tracer gas dosed through the point spreader. The measurement was made in the wake region of a person standing at the fume hood.
Figures 5 and 6 summarise the results of the various measurements in the wake. Figure 5 shows the results with the measurement probe positioned in the wake region of the person standing by the fume hood and with the gas being dosed via the point spreader. Figure 6 is based on measurements with the probe in the same position, but with the gas being dosed through the linear spreader. The dierent cases studied comprise sequences of 10 sash openings for each delay. In Fig. 7 the tracer gas concentrations measured outside the fume hood are expressed as a percentage of the concentration prevailing in the return air when the sash is fully opened. A delay of 2±3 s gave tracer gas concentrations in the wake region of about 10±20% of that in the return air. Delays
exceeding 3 s resulted in concentrations higher than those in the return air from the fume hood. The data in Fig. 7 represent the maximum peak concentrations, while Fig. 8 is based on the average value of the concentrations recorded for each time delay. Dierent values of the delay time are obtained, depending on whether the time is measured from when the sash is fully open or from when it starts to be opened. This means that, to some extent, the time before outward leakage occurs depends on the rate at which the sash is opened. Figure 2, which shows the response with the ventilation control system out of use, shows that the air velocity through the face opening falls to a low level within 0.5±1.0 s. However, even with the control system reacting immediately, the air velocity through the face opening
Fig. 5. Maximum measured concentration during all the test openings for the respective time delays. The measurement sonde was positioned in the wake region and the tracer gas was dosed through the point spreader.
Variable air volume fume hood controllers
149
Fig. 6. Maximum measured concentration during all the test openings for the respective time delays. The measurement sonde was positioned in the wake region and the tracer gas was dosed through the linear spreader.
Fig. 7. Maximum tracer gas concentration in the wake region and in the breathing zone expressed as a percentage of the concentration in the return air from the fume hood.
Fig. 8. Average tracer gas concentration in the wake region and in the breathing zone expressed as a percentage of the concentration in the return air from the fume hood.
150
L. Ekberg and J. Melin
Fig. 9. The mean value of the lowest measured velocity through the face opening.
still drops brie¯y to a low level, although this is of only short duration, as can be seen from Fig. 3. Figure 9 shows the mean value of the lowest measured air velocity for each time delay group. As can be seen from the ®gure, the velocity drops to a low level in all cases. If the time delay is measured not from when the sash starts to be opened or from when it is fully open, but from when the air velocity through the face opening has dropped to a low level in terms of safety, then we have an estimate of the requirements in respect of the response time of the control system, regardless of the rate at which the sash is opened. Also using this de®nition a delay exceeding 3 s will lead to tracer gas concentration peaks in the wake region that are higher than the concentrations prevailing in the return air from the fume hood. CONCLUDING REMARKS
The measurements presented in this paper indicate that a serious loss of the safety function may be observed temporarily if the exhaust air ¯ow rate through a fume hood is not re-established within about 3 s after the sash is moved towards a more open position. The outward leakage associated with a response time of 4 s may be more than 20 times higher than the leakage at a response time of 3 s. Decreasing the delay to 1 s may decrease the outward leakage by another order of magnitude. Indeed, the seriousness of the dierent levels of outward leakage observed depends on the substances being handled in the fume hood (toxicity etc.), but even with a carefully designed and tested fume hood, there is no absolute guarantee against leakage occurring. The measurements were made under virtually ideal conditions. The only departure from the ideal
was the presence of a stationary person in front of the fume hood. Under more typical conditions, the fume hood could be exposed to various other external perturbations, such as the user making various movements as needed for dierent tasks, the presence of equipment in the fume hood which might aect the air ¯ow pattern through the face opening, or draughts in the vicinity of the fume hood. AcknowledgementsÐThis work is supported by The Swedish Council for Building Research. The authors wish to thank Professor Enno Abel and Professor Ove Strindehag for their valuable comments and suggestions. REFERENCES Ahmed, O and Bradley, S. A. (1995) An Approach to Determining the Required Response Time for a VAV Fume Hood Control System, Laboratory HVAC. American Society of Heating Refrigerating and AirConditioning Engineers, Inc., Atlanta, GA, pp. 31±37. ASHRAE Standard (1995) Method of Testing Performance of Laboratory Fume Hoods, ANSI/ASHRAE 110±1995. American Society of Heating, Refrigerating and AirConditioning Engineers, Inc, Atlanta, GA. Lacey, D. R. (1989) Observed Performance of VAV Hood Controls, ASHRAE Transaction, Part 2, pp. 817±824. American Society of Heating Refrigerating and AirConditioning Engineers, Inc, Atlanta, GA. Ljungqvist, B. (1991) Some observations on aerodynamic types of laboratory fume hoods. In Ventilation '91, Proceedings from the Third International Symposium on Ventilation for Contaminant Control, pp. 569±573. American Conference of Governmental Industrial Hygienists, Inc., Cincinnati, OH, USA. Melin, J. (1997) Measurements and analyses of the performance of laboratory fume hoods. Document D40:1997, Department of Building Services Engineering, Chalmers University of Technology, GoÈteborg, Sweden. Rabiah, T. M., Garrison, R. P. and Sachdev, R. K. (1989) Comparison of Variable Volume Fume Hood Controllers, ASHRAE Transaction, Part 2, pp. 837±844. American Society of Heating Refrigerating and Air-Conditioning Engineers, Inc, Atlanta, GA.
Ann. occup. Hyg., Vol. 44, No. 2, pp. 143±150, 2000 # 2000 British Occupational Hygiene Society Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain. 0003±4878/00/$20.00
Required Response Time for Variable Air Volume Fume Hood Controllers LARS E. EKBERG* and JAN MELIN Department of Building Services Engineering, Chalmers University of Technology, S-412 96 GoÈteborg, Sweden This paper describes results from tests made with the aim of investigating how quickly the exhaust air ¯ow rate through fume hoods needs to be controlled in order to prevent contaminants from leaking out of the fume hood and putting the safety of the laboratory personnel at risk. The measurements were made on a laboratory fume hood in a chemical laboratory. There were no other fume hoods in the laboratory, and the measurements were made without interference from persons entering or leaving the laboratory or walking about in it. A tracer gas method was used with the concentration of dinitrogen oxide (N2O) being recorded by a Foxboro Miran 101 infra-red gas analyser. In parallel with the tracer gas measurements, the air velocity through the face opening was also measured, as was the control signal to the damper controlling the air ¯ow rate. The measurements show an increased outward leakage of tracer gas from the fume hood if the air ¯ow rate is not re-established within 1±2 s after the sash is opened. If the delay exceeds 3 s the safety function is temporarily defeated. The measurements were made under virtually ideal conditions. Under more typical conditions, the fume hood could be exposed to various other external perturbations, which means that the control system should re-establish the correct exhaust ¯ow more quickly than indicated by the measurement results obtained under these almost ideal conditions. # 2000 British Occupational Hygiene Society. Published by Elsevier Science Ltd. All rights reserved. Keywords: laboratory ventilation; control system; face velocity; fume hoods
INTRODUCTION
Historically, laboratory fume hoods have usually operated with a constant air ¯ow rate. However, it has become increasingly common in present-day modern laboratories to control the exhaust air ¯ow rate from fume hoods in order to maintain a constant air velocity through the face, regardless of the position of the sash. There are two main methods that are normally used for this purpose: one measures the air velocity, while the other responds to the position of the sash, with the correct ¯ow rate being set from one of these parameters. When using the face velocity as a control parameter, there is generally some form of velocity sensor ®tted to the inside of the fume hood. However, there are a number of factors that can aect the Received 8 December 1998; in ®nal form 12 July 1999. *Author to whom correspondence should be addressed. Tel.: +46-31-772-1162; fax: +46-31-772-1152; E-mail: [email protected] 143
measured velocity so that it is not representative of the average velocity through the face opening. Examples of such factors include equipment inside the fume hood upsetting the ¯ow pattern around the sensor, the creation of draughts within the hood due to the movements of the operator, or a draught across the face opening. One further important condition to consider when measuring the air velocity close to the fume hood wall is that the air velocity through the face opening is not necessarily always uniformly distributed. Results from air velocity measurements, carried out using a hot wire anemometer at several points for three dierent face openings in a fume hood installed in a laboratory, reveal that there may be positions where the measurement does not correspond fully with the average face velocity. This means that there may be a need to correct the measured air velocity in order to get a value representative of the average face velocity. This eect can be quite pronounced in measurement positions close to the edge of the sash and along the sides
144
L. Ekberg and J. Melin
and the lower edge of the fume hood opening, where also substantial velocity ¯uctuations due to turbulence can be expected. Using the position of the sash as the control parameter employs the relationship that the area of the face opening is proportional to the position of the sash. In purely practical terms, this is usually sensed by securing a wire to the sash to drive a potentiometer, the signal from which then controls a damper or a fan to provide the correct ¯ow rate. An important dierence between the two control principles is their speed of operation. In papers by Rabiah et al. (1989) and Lacey (1989), results are presented from trials where the response times for dierent types of commercially available control systems were tested. The results show that systems that employ the sash position as the control parameter generally are faster in their response than systems that use the air velocity as the control parameter. There is no doubt that, in almost all cases, it is an advantage to control the exhaust air ¯ow from the fume hood. Similarly, there is no doubt that dierent types of control systems react at dierent speeds. How quickly does the ¯ow need to be controlled to ensure that the fume hood provides adequate protection? A study by Ahmed and Bradley (1995) describes measurements intended to investigate the necessary response time for fume hood control systems. The measurements were based on ANSI/ASHRAE Standard 110-1995, using sulphur hexa¯uoride as the tracer gas and dosing it into the fume hood via an ASHRAE Standard tracer gas ejector, with the measurement point positioned at the breathing zone of a dummy. The experiments were carried out under ideal conditions in two laboratories on fume hoods without existing ventilation control. Instead, ¯ow control was eected by manual control of the fan motor speed. A portable electron capture cell detector was used for analysis of the gas concentration, recording the results in one laboratory on a strip chart recorder and in the other on a PC. Opening the sash from 25% to fully open, without increasing the ventilation ¯ow rate, reduced the air velocity through the face opening from 100 fpm (0.5 m sÿ1) to 25 fpm (0.13 m sÿ1). The time from the sash reaching fully open until tracer gas was recorded varied over a wide interval. It was shown that no gas leaked out before about 2±3 s from the sash being fully opened. In another series of measurements, the sash was opened from 25% to fully open and, after a certain time, the speed of the fan was increased to restore the average front velocity of 100 fpm. In this case, the time between the sash being fully opened and the instance when the speed of the fan needs to be
increased in order to avoid outward leakage was of the order of 2±3 s. The present paper describes results from another investigation made with the aim of investigating how quickly the ventilation air ¯ow rate needs to be controlled in order to prevent contaminants from leaking out of the fume hood and putting the safety of the laboratory personnel at risk. EXPERIMENTAL PROCEDURE
The measurements were made on a laboratory fume hood of the type shown in Fig. 1. There were no other fume hoods in the laboratory, and the measurements were made without interference from persons entering or leaving the laboratory or walking about in it. No equipment was placed in the fume hood. The fume hood had an aperture width of 102.5 cm and a fully open face height of 74 cm. The sash was intentionally designed so that it could not be completely closed: when pulled down as far as possible, a gap was left at the bottom in order to maintain a basic through ¯ow rate of 100 m3 hÿ1. The fume hood had a control system for maintaining a constant average face velocity of 0.5 m sÿ1, regardless of the position of the sash. The control system was of the sash position sensing type, with the control signal from the potentiometer that sensed the sash position controlling a valve for adjustment of the air ¯ow rate. Moving the sash to
Fig. 1. Sketch of the aerodynamic type of laboratory fume hood studied.
Variable air volume fume hood controllers
its fully open position resulted in an increase in the air ¯ow rate towards 1365 m3 hÿ1. The tracer gas method was used in order to determine how quickly the exhaust air ¯ow rate needed to be regulated. Nitrous oxide, N2O, was used as the tracer gas, with the concentration being recorded by a Foxboro Miran 101 IR gas analyser. According to the manufacturer, the detection limit for N20 using this instrument is 0.5 ppm. The measurements were made with a person standing in front of the fume hood, making no movements apart from opening the sash. The measurement probe was placed in two positions: in the wake region in front of the person standing by the fume hood, 3 cm above the airfoil and 1 cm from the plane of the sash, and in the person's breathing zone, 160 cm above the ¯oor and 5 cm from the plane of the sash. The time taken until the tracer gas was registered by the gas analyser was corrected to allow for the instrument's dead time, that is the time it took for the gas to travel from the measurement probe to the instrument. The dead time was found by using the Miran 101 gas analyser together with another instrument that has a negligible inertia (Melin, 1997). Simultaneous measurements with these two instruments in a duct into which tracer gas was dosed revealed that the dead time was 1.5 s. The tracer gas was dosed through a linear spreader and, in certain cases, through a point spreader. Both the point spreader and the linear spreader were used for the measurements in the wake region of the person standing in front of the fume hood. The point spreader was positioned 15 cm into the fume hood, at a height of 25 cm above the working surface. The line spreader was also placed 15 cm into the fume hood, but at a height of 10 cm above the working surface. For the measurements in the breathing zone, only the linear spreader was used, being positioned 15 cm into the fume hood and at a height of 37.5 cm above the working surface. The tracer gas ¯ow rate through the point spreader was 87 l. hÿ1, while that through the linear spreader was 120 l. hÿ1. The gas ¯ow was measured using a rotameter. In parallel with the tracer gas measurements, the air velocity through the face opening was also measured, as was the control signal to the control valve. The face velocity was measured using an air velocity sensor incorporating a thermistor, positioned 4 cm below the edge of the sash. With the time constant used for the tests, the response time of this instrument was 0.2 s. A switch was ®tted to the sash in order to be able to measure exactly when it opened. The signals from the gas analyser, the air velocity sensor, the control device and the switch were scanned at a rate of 10 values sÿ1 and stored in a computer. The ®rst stage of the measurements involved investigating the speed of response of the existing
145
control system. In order to estimate the time for the control valve and exhaust air ¯ow rate to change to the required level after a change in the position of the sash, measurements were made that involved quickly opening the sash while logging the control signal and the air velocity through the sash opening. These measurements were made from the closed sash position to three opening heights: onethird open (1/3), two-thirds open (2/3) and fully open (1/1). The sash was opened at a rate of about 0.5 s from the closed to the fully open position. The next part of the investigation was designed for determination of how quickly the air ¯ow through the fume hood needed to be controlled in order to prevent tracer gas from leaking out. For these measurements, the tracer gas was dosed inside the fume hood, and the sash was opened from the closed position to the fully open position at a rate equivalent to 2 s. The reason for taking such a long time to open the sash was to avoid the actual act of opening the sash from creating turbulence in the face opening and thus causing an outward leakage of tracer gas. During the ®rst series of measurements, the control system was completely disconnected, that is there was only the basic ventilation ¯ow through the fume hood. The purpose of these measurements was to determine the time it took for the tracer gas to leak out from the fume hood after the sash was opened. Repeated measurements were made, using both the point spreader and the linear spreader, and with the measurement probe in the wake region of the person standing at the fume hood. In order to prevent a build-up of tracer gas in the laboratory when making the measurements, the control system was activated as soon as the gas analyser had detected the gas. The existing control system was in operation for the second series of measurements. These measurements were intended to determine how quickly the system needed to react and set the correct air ¯ow rate in order to avoid outward leakage from the fume hood. During the ®rst set of measurements in this series, the control system was allowed to respond in its normal manner. The sash was repeatedly opened and the tracer gas concentration at the measurement point was recorded. During the next set of measurements, the control system was deactivated for 1 s, that is a delay of 1 s was introduced before the system was allowed to respond and adjust the air ¯ow in the normal way. Again, the sash was opened repeatedly and the tracer gas concentration recorded in the same way as before. The same overall procedure was then followed, but with respective delays of 2, 3, 4, 5 and 6 s. The described procedures were carried out using both point spreaders and linear spreaders for the measurement points in the wake region of the operator and in the breathing zone.
146
L. Ekberg and J. Melin Table 1. Delay times for the existing control system
Sash opening
1/3 2/3 1/1
Sash position command
Face velocity
Delay (s)
Standard deviation (N = 15)
Delay (s)
Standard deviation (N = 15)
0.9 1.4 1.8
0.07 0.17 0.07
1.7 1.7 1.9
0.31 0.15 0.16
RESULTS AND DISCUSSION
Existing control system Table 1 shows the delay times that could be determined from the measurements carried out with the existing control system in operation. The delay time for the sash position command is de®ned as the time from when the sash was opened until the control signal had reached a stable level. The delay time for the face velocity is de®ned as the time from when the sash was opened until the air velocity at the measurement point had ®rst reached 0.5 m sÿ1. The table shows that, for the worst case of when the sash was opened to its fully-open position, the air velocity through the face opening responded in less than 2 s from starting to open the sash. Operation without air¯ow control Figure 2 is an example of a test, showing how the air velocity drops and then stabilises at a low velocity when there is no increase in the exhaust air ¯ow rate due to the control system having been disengaged. A total of 10 measurements of the same type were carried out each with a linear spreader and a point spreader in use. The results are discussed in the following. The time taken for the tracer gas to reach the instrument varied somewhat, depending on
whether the gas was dosed through the point spreader or through the linear spreader. When dosing the gas through the point spreader, it was found that the average value of the time between the sash being fully opened and the gas reaching the instrument was 2.6 s, with a standard deviation of 0.9 (N = 10). The shortest time was 1.2 s and the longest time was 4.2 s. When dosing the gas through the linear spreader, it was found that the average value of the time, as de®ned above, was 3.0 s, with a standard deviation of 0.9 (N = 10). In this case, the shortest time was 1.7 s and the longest time was 4.6 s. The measured results thus provide an average value for all the measurements (that is with both point spreader and linear spreader) of the time from the sash being fully opened until the tracer gas had reached the instrument, of 2.8 s, with a standard deviation of 0.9. The times noted above are de®ned as being the time from when the sash reached its fully open position until the tracer gas reached the measurement point. Alternatively, it could be de®ned as the time between starting to open the sash and the gas reaching the measurement point. By this de®nition, the average value for both the point spreader and the linear spreader would be 4.8 s, with a standard deviation of 0.9. The fact that an outward leakage of the tracer
Fig. 2. Air velocity in response to opening the sash with the control system out of use.
Variable air volume fume hood controllers
gas occurs is naturally due to the eect of the average velocity through the face opening falling to such a low value that it is insucient to prevent undesired spread of contaminants. Ljungqvist (1991) showed that reducing the air velocity through the face opening by 20±40% of the nominal value of 0.5 m sÿ1 results in an increased outward leakage. Figure 2 shows how the air velocity through the face opening falls when the sash is opened without increasing the air ¯ow rate. After only 0.5±1.0 s from starting to open the sash, the velocity has dropped by about a factor of two, and consequently, there is a risk of increased outward leakage. The results from measurements with the control system completely out of operation and measuring the time until tracer gas is detected, indicate how quickly a control system needs to respond and start to adjust the air ¯ow rate in order to avoid outward leakage of contaminants. However, further tracer gas measurements carried out at various speeds of the air ¯ow rate control are needed in order to obtain more detailed information about what demands should be made on the control system. Operation with dierent time delays of the control signal This section gives the results from measurements carried out with the existing control system in operation. These measurements were intended to determine how quickly the system needed to react and set the correct air ¯ow rate in order to minimise the outward leakage from the fume hood. During the ®rst set of measurements in this series, the control system was allowed to respond in its normal manner. The sash was repeatedly opened and the tracer
147
gas concentration at the measurement point was recorded. As mentioned previously, the sash was moved from the closed to the fully open position in 2 s. Figure 3 shows the response for the case with no time delay before the control system was allowed to start operating. During the next set of measurements, the control system was deactivated for 1, 2, 3, 4, 5 and 6 s. These measurements were performed in sequences taking about a minute. This meant that, if and when leakage occurred, any tracer gas that had leaked out was evacuated from the laboratory before the next sequence started, thus preventing any build-up of concentration of tracer gas in the laboratory. Figure 4 shows an example of the sequence of changes in the sash position. The ®gure shows the response with a 4-s delay before the control system was activated. Furthermore, it can be shown that the delay of the control signal is approximately equal to the delay between the sash being fully opened and the air velocity being reestablished at its nominal value of 0.5 m sÿ1. Figure 4 shows that, with a 4-s delay in the control system, the tracer gas concentration in the wake region of a person standing in front of the fume hood may reach a few hundred ppm. For comparison it should be noted that the concentration was less than one hundred ppm in the return air from the fume hood when the sash was fully open. Moreover, the measurements in the breathing zone also show a considerable outward leakage of tracer gas with a control system delay of 6 s, that is if there is a delay of 6 s from the sash reaching the fully open position (or 8 s from the time of starting to move the sash) before the correct air velocity is re-established, there will be a considerable outward leakage.
Fig. 3. Control signal and air velocity when opening the sash from closed to fully open over a period of 2 s, with no delay in the control system.
148
L. Ekberg and J. Melin
Fig. 4. An example of a series of measurements of a 4-s delay of the control system and with the tracer gas dosed through the point spreader. The measurement was made in the wake region of a person standing at the fume hood.
Figures 5 and 6 summarise the results of the various measurements in the wake. Figure 5 shows the results with the measurement probe positioned in the wake region of the person standing by the fume hood and with the gas being dosed via the point spreader. Figure 6 is based on measurements with the probe in the same position, but with the gas being dosed through the linear spreader. The dierent cases studied comprise sequences of 10 sash openings for each delay. In Fig. 7 the tracer gas concentrations measured outside the fume hood are expressed as a percentage of the concentration prevailing in the return air when the sash is fully opened. A delay of 2±3 s gave tracer gas concentrations in the wake region of about 10±20% of that in the return air. Delays
exceeding 3 s resulted in concentrations higher than those in the return air from the fume hood. The data in Fig. 7 represent the maximum peak concentrations, while Fig. 8 is based on the average value of the concentrations recorded for each time delay. Dierent values of the delay time are obtained, depending on whether the time is measured from when the sash is fully open or from when it starts to be opened. This means that, to some extent, the time before outward leakage occurs depends on the rate at which the sash is opened. Figure 2, which shows the response with the ventilation control system out of use, shows that the air velocity through the face opening falls to a low level within 0.5±1.0 s. However, even with the control system reacting immediately, the air velocity through the face opening
Fig. 5. Maximum measured concentration during all the test openings for the respective time delays. The measurement sonde was positioned in the wake region and the tracer gas was dosed through the point spreader.
Variable air volume fume hood controllers
149
Fig. 6. Maximum measured concentration during all the test openings for the respective time delays. The measurement sonde was positioned in the wake region and the tracer gas was dosed through the linear spreader.
Fig. 7. Maximum tracer gas concentration in the wake region and in the breathing zone expressed as a percentage of the concentration in the return air from the fume hood.
Fig. 8. Average tracer gas concentration in the wake region and in the breathing zone expressed as a percentage of the concentration in the return air from the fume hood.
150
L. Ekberg and J. Melin
Fig. 9. The mean value of the lowest measured velocity through the face opening.
still drops brie¯y to a low level, although this is of only short duration, as can be seen from Fig. 3. Figure 9 shows the mean value of the lowest measured air velocity for each time delay group. As can be seen from the ®gure, the velocity drops to a low level in all cases. If the time delay is measured not from when the sash starts to be opened or from when it is fully open, but from when the air velocity through the face opening has dropped to a low level in terms of safety, then we have an estimate of the requirements in respect of the response time of the control system, regardless of the rate at which the sash is opened. Also using this de®nition a delay exceeding 3 s will lead to tracer gas concentration peaks in the wake region that are higher than the concentrations prevailing in the return air from the fume hood. CONCLUDING REMARKS
The measurements presented in this paper indicate that a serious loss of the safety function may be observed temporarily if the exhaust air ¯ow rate through a fume hood is not re-established within about 3 s after the sash is moved towards a more open position. The outward leakage associated with a response time of 4 s may be more than 20 times higher than the leakage at a response time of 3 s. Decreasing the delay to 1 s may decrease the outward leakage by another order of magnitude. Indeed, the seriousness of the dierent levels of outward leakage observed depends on the substances being handled in the fume hood (toxicity etc.), but even with a carefully designed and tested fume hood, there is no absolute guarantee against leakage occurring. The measurements were made under virtually ideal conditions. The only departure from the ideal
was the presence of a stationary person in front of the fume hood. Under more typical conditions, the fume hood could be exposed to various other external perturbations, such as the user making various movements as needed for dierent tasks, the presence of equipment in the fume hood which might aect the air ¯ow pattern through the face opening, or draughts in the vicinity of the fume hood. AcknowledgementsÐThis work is supported by The Swedish Council for Building Research. The authors wish to thank Professor Enno Abel and Professor Ove Strindehag for their valuable comments and suggestions. REFERENCES Ahmed, O and Bradley, S. A. (1995) An Approach to Determining the Required Response Time for a VAV Fume Hood Control System, Laboratory HVAC. American Society of Heating Refrigerating and AirConditioning Engineers, Inc., Atlanta, GA, pp. 31±37. ASHRAE Standard (1995) Method of Testing Performance of Laboratory Fume Hoods, ANSI/ASHRAE 110±1995. American Society of Heating, Refrigerating and AirConditioning Engineers, Inc, Atlanta, GA. Lacey, D. R. (1989) Observed Performance of VAV Hood Controls, ASHRAE Transaction, Part 2, pp. 817±824. American Society of Heating Refrigerating and AirConditioning Engineers, Inc, Atlanta, GA. Ljungqvist, B. (1991) Some observations on aerodynamic types of laboratory fume hoods. In Ventilation '91, Proceedings from the Third International Symposium on Ventilation for Contaminant Control, pp. 569±573. American Conference of Governmental Industrial Hygienists, Inc., Cincinnati, OH, USA. Melin, J. (1997) Measurements and analyses of the performance of laboratory fume hoods. Document D40:1997, Department of Building Services Engineering, Chalmers University of Technology, GoÈteborg, Sweden. Rabiah, T. M., Garrison, R. P. and Sachdev, R. K. (1989) Comparison of Variable Volume Fume Hood Controllers, ASHRAE Transaction, Part 2, pp. 837±844. American Society of Heating Refrigerating and Air-Conditioning Engineers, Inc, Atlanta, GA.