Stack Effect in a Glazed Stairwell
Vital Signs Case Study Report
Randolph Fritz & Pete Keller
December 6, 1999

Introduction

Intent

The "stack effect" is a physical property of airflow in a vertical space vented at top and bottom. It occurs when cool air at the bottom is warmed. The warm air rises escapes. The air will then be replaced by air being drawn in at the bottom. This effect can be used to naturally ventilate a building.

As the science of building becomes ecologically aware, many of these more subtle effects will be more closely examined. Is the stack effect powerful enough to be investing building resources in the concept? The answer to that will take many case studies of many different buildings in many places. Our intent here is to provide one of these studies. We would like to show whether a stack effect is measurable in a chosen space.

The stack effect suggests that air movement will correlate to the difference in temperature between the bottom and the top of the space. A large temperature difference should be accompanied by a large amount of airflow. Because the glazing heats the air inside the space, we expected a larger temperature difference on warm days than on cool days.

The place

Eugene Oregon, at the south end of the Willamette River valley of the Pacific Northwest of the USA, has a mild climate with a warm mild summer and a cool gray rainy winter. A through set of maps of the area can be found at [LCATLAS].

A more complete climate summary can be found at [LCDS].

The space

	Third-floor landing,  looking down Second floor landing, looking up Second floor landing, looking down First-second floor landing, looking up

A stairwell in the Lawrence Hall of the University of Oregon in Eugene, Oregon, USA. The space has dramatic thermal properties.

• The glazing, comprising an estimated ³/₄ of the walls, takes in a large amount of solar radiation.
• The glazing, being single, also conducts large amounts of heat in and out of the building.
• The massive concrete stair acts as a thermal mass.
• There is a radiator on the ground floor of the stairwell.
• The stairwell is isolated from the rest of the building by fire doors.

Hypothesis

The air movement in the stairwell will correlate with the temperature difference between the top and bottom of the space.

Methodology

Overview

In order to determine whether a stack effect occurs, we studied whether temperature can be correlated with airflow through the space. If there was, in fact, a stack effect, expected to see a rise in air flow corresponding to a rise in the difference between indoor and outdoor temperatures.

To determine this we took temperature readings at the top and the bottom of the space, and also measured the velocity of the air coming into the space underneath the doors, again at the top and bottom of the space.

Instruments

All these instruments are described in detail in [VSTK].

• A thermometer. We used the dry-bulb thermometer of a Taylor sling psychrometer.
• A Solomat 429 hot-wire anemometer. This extremely sensitive instrument measures the drop in temperature from airflow past it.
• A Borozin smoke gun, also called a smoke puffer or powder puffer; a small hand pump which sprays zinc-white dust, used for determining the direction of drafts.

Execution

We began by leaving the, doors propped open while we measured but this proved to be too influenced by the building's HVAC, and also proved vulnerable to the foot traffic in the space. We got more consistent readings with the doors shut, measuring airflow underneath the doors.

We found that the hot-wire anemometer was sensitive enough to be affected by wind gusts, doors opening and closing, and even the operator's breath. The device must be observed over a period of time, and wind gust effects must be reconciled through human judgment. For our purposes, we probably would have been better off with a logging anemometer.

The heater at the bottom of the stairs was turned on during our measurements.

Towards the end of our measuring, we realized that we'd get more useful air velocity data by recording an error range; we used that as a basis for error ranges in earlier data.

We only found out about the Borozin smoke gun after we had done the main data gathering, and only had time for one set of measurements with it.

Data

Main measurements

Date and time	Ground floor		Second floor landing		Third floor landing
	Temp. ºF	Air Velocity ft/min.	Temp. ºF	Air Velocity ft/min.	Temp. ºF	Air Velocity ft/min.
Fri. 11/12/99 1900 (note 1,6)	75	200±31	70	40±4	72	120±6
Sat. 11/13/99 1515 (note 1,6)	74	125±20	73	10±1	72	43±2
Sun. 11/14/99 1300 (note 3,6)	70	120±19	71	10±1	71	40±2
Mon. 11/15/99 1500 (note 6)	70	50±8	70	50±5	72	35±2
Mon. 11/15/99 1645 (note 6)	72	90±14	71	60±6	71	15±1
Tue. 11/16/99 1900 (note 6)	69	(note 4)	64	90±10	67	70±4
Wed. 11/17/99 1135 (note 5,7)	69	165±35	66	60±10	66	180±5
Thu. 11/18/99 1200 (note 7)	61	177.5±7.5	60	70±10	61	58.5±4.5
Thu. 11/18/99 1330 (note 7)	60	46±10	63	123±1	65	37±2

Notes

1. Measured with the ground floor and third floor doors propped
2. Passerby closed ground floor door during measurement.
3. Measured under door; this rule was followed for all succeeding measurements.
4. Data confused by door opening & closing in corridor
5. Finally worked out how to deal with meter variations--noting a range of variation.
6. Error estimated from the average percent error of the last three error ranges in the column, which were noted by the operator.
7. Error as noted by operator.

Additional measurements

Reported weather

The following table covers the weather information from the National Climatic Data Center's Surface Weather data set [SURW]. Measurements are taken from the table entry which is closest in time. Centigrade temperatures have been converted to Fahrenheit and rounded to the nearest whole degree. [METAR] is helpful in interpreting the tables.

Convection against windows.

On the landing between the second and third floors, we took two measurements of airflow against the windows.

	Date Time	Inside temp. (°F)	Outside temp. (°F)	Air velocity, north window (ft./min)	Air velocity, south window
	Fri 1900	70	63	40	40
	Sat 1515	73	72	0	10

Air flow was, in both cases, down.

Direction of air flow

Study of dust patterns on the door at 16:29 on 23 November (outside temperature approx. 50°F) revealed airflow, and we followed up with tests with the Borozin smoke gun.

Apparently the dust accumulates on the side of the door which takes in air, rather as dust accumulates on the intake grill of an air duct.

Leaks

• The mullions and muntins clearly leak--one can see trails where water has brought in dust and then dried.
• The doors are four-hour fire doors, and apparently mostly leak at the thresholds.

Geometry of the space

Analysis

The staircase apparently "breathes" with the building. Depending on outside and inside temperatures, air flows differently at different times.

• On 24 November, when I checked the direction of flow, apparently an actual stack effect was running, with small amounts of air being pulled in under the downstairs door and a substantial amount of air being pushed out of the third-floor door. It seems likely that air was being drawn into the staircase from outside.
• At no time did the measured flows add up to zero. Apparently air is leaving and entering the space through leaks around the mullions.

Mathematics

According to [MEEB] pages 127-8, the airflow through the space should be approximately:
[ed. CAUTION! ERROR IN THIS EQUATION. Please refer to ASHRAE Handbook, 2005 Fundamentals, equation 27.30.] 

or , dividing by A

Where:

Q

is the airflow in ft³/min or m³/sec

C

is a constant, equal to 313 ft^0.5/min or 91.1 m^0.5/sec

A

is the area of the stack or vents (whichever is smaller) in ft² or m²

h

is the height in feet or meters

t_i and t_o

are the inside and outside temperatures in °F, respectively.

v

is air velocity in ft/min or m/sec

This formula, however, is pretty plainly an approximation. Apart from the question of what happens when t_o > t_i (when the airflow would become imaginary; the value of Q or v explodes when t_i goes to zero. Most likely the formula is most accurate around a particular temperature, probably 72°.

First fit

Graphing air velocity against where t_i is taken from measured inside temperature and t_o is taken from reported outside temperature produces the a chart which does not show the relationship:

v

Second fit

But, graphing air velocity against the difference between top and bottom temperature in the space, at least suggests a relationship.

v

Third fit

After a lot of thought and analysis, we decided that:

• The temperature at the top of the stairs was perhaps a more accurate reflection of the temperature outside that stairwell than the measurement taken at the airport
• There was probably a base level of air flow in the space, perhaps driven by the building HVAC.

With that in mind, we produced the following chart, which fits all the data very well except for one point:

v

Conclusions

As [MEEB] pages 127-8 say, when a sufficient temperature differential builds up, there is a stack effect in the space. However, the temperature differential in the time period we were measuring in seldom was large enough to drive the effect. Surprisingly, though, the reported exterior temperature does not seem to correlate at all with the stack effect. It may be:

1. That the temperature at Eugene Airport (Mahlon-Sweet Airport, to be technical) is not very close to the temperature on the University of Oregon's Eugene campus.
2. Characteristics of the space may make the measured outside temperature insignificant.
3. Another, as yet undetermined, factor has come into play.

We have two further hypotheses covering this:

Hypothesis 1

It is possible that the vertical air circulation of the stack effect is governed by the same principles as horizontal convection. Perhaps the term "stack effect" is misleading in this way, suggesting that the air movement is caused by the vertical property of the space. It seems likely that any lengthy space, whether vertical or horizontal, will display the same convection properties, it is a three degree difference in temperature that is required to move air. This makes physical sense because gravity is a weak force, and the wait for the air to move because of the effects of gravity will be interrupted by the air being pushed around by horizontal density differences. The wind doesn't blow much, up and down.

Perhaps, then, it is unwise to try to design a stack effect into a building, and much better to try to connect a naturally warm outdoor area to a naturally cool outdoor area, wherever these two locations are on the horizontal or vertical planes. A vertical space might be used to do this, but it would function no better than a horizontal space that fulfills the requirement of a three degree temperature difference.

Hypothesis 2

It may be that this space simply doesn't usually get warm enough to produce a substantial stack effect in the fall. In addtion, it seems likely that the thermal masses of staircases and walls actually reduce the stack effect, since they tend to moderate temperature differences. And the glazing is both thin and leaky. It may be that the stack effect is not actually of all that much use in an Oregon fall, or it may be that stacks have to be better designed to take advantage of it. A space designed to effectively produce the stack effect would ideally be better insulated and have either minimum thermal mass or most thermal mass insulated.

References

[LCATLAS] Lane County Atlas, University of Oregon InfoGraphics Lab, Spring 1998, Winter 1999, ongoing.

[LCDS] Local Climate Data Summaries, Western Regional Climate Center, consulted 12/5/99.

[MEEB] Mechanical and Electrical Equipment for Buildings, 8th Ed., Benjamin Stein, John Reynolds. ISBN: 0-471-52502-2. John Wiley & Sons, Inc., 1992. New York, USA; Chichester, UK; Weinheim, Germany; Brisbane, Australia; Singapore; Toronto, Canada.

[METAR] Surface METAR Observations Users Manual, US Navy, Fleet Numerical Meteorology and Oceanography Detachment, Asheville, NC.

"On the way," preliminary papers and reports on this study. Randolph Fritz and Pete Keller, Fall 1999. Research Proposal, Methodology, Preliminary Results and Analysis.

[SURW] Surface Weather Observations from KEUG, National Climate Data Center, freely available to educational users (.EDU domains) all others pay cash.

[VSP] Vital Signs Project, Cris Benton, Principal Investigator. Center for Environmental Design, University of California at Berkeley, Berkeley, CA, USA. Initiated 1992, ongoing.

[VSTK] "Vital Signs Toolkit Roster," Vital Signs Project, June 1998. PDF format.

Acknowledgments

Professor Walter Grondzik, for advice and finding the Borozin smoke gun.

Assistant Professor Alison Kwok, former research assistant to Cris Benton, principal investigator of the Vital Signs Project, without whom the subject would be quite different.