Electric and gas-fueled clothes dryers continue to be associated with
a large number of fire incidents in the U.S. In 1998, an estimated 15,600
clothes dryer fire incidents occurred, resulting in 20 deaths, 370 injuries,
and U.S. $75.4 million in property damage .
With approximately 80 million clothes
dryers in use in the U.S.  and an average product life of about
13 years, the potential for future fire incidents exists. The operation
of an electric clothes dryer involves unattended operation, high
temperatures, high voltages and currents, and a potential fuel
source (i.e., the materials being dried). Since these conditions
are intrinsic to clothes dryer operation, any attempts to reduce
the consumer risk associated with this product must take these
factors into account.
following is an edited version of a paper delivered
at the 55th Annual International Appliance Technical
Conference (IATC), held March 29-31, 2004 in Lexington,
paper’s authors were awarded the Dana Chase,
Sr. Memorial Award for the best paper presented at
Prior studies by the U.S. Consumer Product Safety Commission (CPSC) staff
 and The National Fire Prevention Association (NFPA) have documented
the scenarios surrounding reported clothes dryer fires. The reports showed
that elevated temperatures could be present in a clothes dryer with no
warnings directed at the consumer.
CPSC researched how new and emerging sensor technologies could be used
to reduce hazards associated with consumer products. The study identified
a variety of sensor types with the potential to prevent, correct, or alert
the consumer of an incipient hazard condition .
A research project was undertaken to examine how sensor technology could
be applied to a clothes dryer to reduce the risk of fire. This project
had three primary objectives to demonstrate the following: the application
of new and emerging sensor technologies can be a means of detecting abnormal
operating conditions that might lead to a risk of fire; the use of multiple
sensor inputs of different types can be used to distinguish specific abnormal
operating conditions; and condition-based monitoring can be used to assess
clothes dryer operation outside of normal ranges.
Figure 1: Normal Operation Temperature Profile. CLICK to
see the full-size image.
For this project, an electric clothes dryer was instrumented with a variety
of sensors. The dryer was operated under normal and abnormal conditions,
during which time the sensor outputs were recorded. Sensor data were examined
to establish the normal operating characteristics of the clothes dryer,
and to determine if an abnormal operating scenario or a condition that
could lead to an abnormal operating scenario could be discriminated prior
to its creating a potential consumer hazard.
The clothes dryer used in these experiments is a typical retail model.
The appliance consists of a rotating tumbler, a removable lint screen,
an electric heating element, and a blower that creates negative pressure
in the tumbler as it exhausts the heated and moistened air out the exhaust
ducting. The model clothes dryer used has a lint screen accessible from
the top and located in the rear of the appliance, not on the dryer door.
The clothes dryer is equipped with two auto-reset temperature-limiting
controls. The first device is the control thermostat, positioned just after
the blower exhaust. During normal operation, this thermostat cuts electric
power to the heating element once the exhaust airflow temperature reaches
a preset level. When the airflow has cooled, the thermostat resets and
re-energizes the heating element. The second device is a high-limit thermostat,
positioned at the heating element air intake. A temperature increase at
the high-limit thermostat for any reason (such as a reduced airflow) will
activate the thermostat and de-energize the element. This thermostat also
re-energizes the heating element after cooling down.
The following nine sensing technologies were integrated into the clothes
Temperature Sensor—Temperatures were monitored by sensors positioned
to detect the following: heating element intake air, tumbler intake air,
control thermostat (blower exhaust air), tumbler surface, exhaust duct
air, control panel interior air, clothes dryer, interior air outside the
tumbler, and ambient room air.
Infrared Sensor (IR)—An IR sensor was installed in the stationary
rear wall of the tumbler to monitor the temperature of the load being dried.
Humidity Sensor—The relative humidity of the exhaust air, the clothes
dryer interior (outside the tumbler), and the room ambient air were monitored.
Rotation—Tumbler rotation was monitored with a Hall Effect sensor.
Thirty-two magnets were attached to the tumbler rim. As the magnets passed
by the sensor, it would emit voltage pulses. The rotation rate of the tumbler
was calculated from the number of pulses observed during a fixed time interval.
Pressure Sensor—The air pressure drop across the lint screen was
measured with a solid-state, differential pressure sensor capable of detecting
very small pressure changes.
Vibration Sensor—The vibration of the clothes dryer’s motor
was recorded with two single-axis accelerometers mounted perpendicularly
on the motor mount. The accelerometer outputs were sampled 128 times a
sec. A Fast-Fourier Transform (FFT) was performed on the data to generate
a power spectrum with 0.125-Hz resolution and a 64-Hz bandwidth.
Current Sensor—The electric currents drawn by the motor and the heating
element were monitored separately with toroidal current sensors.
air concentrations of various gases were measured with three sensors. Carbon
monoxide (CO) and carbon dioxide (CO2)
were sensed separately. Volatile organic compounds (VOC) were sensed with
broad-spectrum sensor located in the dryer exhaust. This device responds
to many organic complexes, alcohols, and chlorinated compounds.
Airflow Sensor—Airflow in the exhaust was measured using a hotwire
anemometer. Data collection was performed with personal computers connected
to multi-input data acquisition modules. Spreadsheet software was used
to perform the calculations (averages, slopes, durations, etc.) and to
generate graphs of the recorded information. Graphical icons represent
sensors that are linked to storage files and displays. The dials, gauges,
and charts displayed data in real-time as the load was dried.
Figure 2: Pressure Change During a Drying Cycle. CLICK to
see the full-size image.
A standard test load of eight all-cotton bath towels was chosen for the
testing. The load was washed in a household clothes washer before drying
in the instrumented clothes dryer. The load was dried for 60 min, the time
necessary to fully dry the towels.
Both “normal” and “abnormal” tests were conducted.
Normal testing was defined as ordinary operation without any modifications
to the clothes dryer. Abnormal testing involved modifying the clothes dryer
or the load to simulate conditions associated with lack of maintenance,
misuse, improper installation, component failure, or end-of-life circumstances.
The following nine types of abnormal operating conditions were defined
for this testing:
Overfilled Tumbler—A wet load of 15 or 20 towels was placed in the
tumbler, and the clothes dryer was operated for 60 min.
Modified Electric Heating Element—A partial short-circuit was created
in the heating element to increase power dissipation by 60 percent.
Blocked Lint Screen—Lint screens with 25 percent, 50 percent, 75
percent, or 100 percent of the screen area covered were installed in the
Blocked Exhaust Duct—An orifice was installed in the exhaust ducting,
and restrictions of 25 percent, 50 percent, 75 percent, and 100 percent
of the cross-section of the duct were created.
Air-Leak Gaps—Air-leak gaps of 1/4 to 1/2 in were created in the
airflow path of the clothes dryer. The gaps were created between the dryer
exhaust and the exhaust ducting at the blower intake housing, at the ducting
behind the tumbler and before the lint screen, and at the gasket between
the tumbler and the stationary dryer wall.
Combustible Vapors—Samples of volatile, flammable chemicals were
added to wet or dry towel loads, which were then dried.
Smoldering Combustion—Sample materials (cotton rope) with non-flaming
combustion were tumbled separately in the clothes dryer or added to a dry
Flaming Combustion—Samples of flaming cotton towel were tumbled separately
in the clothes dryer.
Spontaneous Combustion—Soybean or linseed oil was added to a towel
load, which was then dried and left in the tumbler after stopping the dryer
before its cooldown cycle.
For the combustion tests, the door of the clothes dryer was replaced with
a plastic shield.
During the testing, all the sensor inputs were recorded. The data were
examined to determine signal responses during normal testing and their
deviations from the normal responses during the abnormal tests. Not every
sensor’s response changed from its nominal value when an abnormal
test was executed. Only those sensors with the most significant normal
to abnormal variations were considered.
Figure 3: Tumbler Intake Air Temperatures. CLICK to
see the full-size image.
Nominal sensor responses to clothes dryer operation were established
to fully characterize the appliance’s operation and to provide
a measure against which abnormal test sensor outputs could be compared.
The cycling of the control thermostat modulates air temperatures in the
tumbler. As a consequence, many other measured temperatures track the air
temperature around the control thermostat. During normal operation, the
air temperature at the control thermostat rose to about 80°C. Then,
the thermostat activated and de-energized the heating element. Once the
air temperature cooled to about 55°C, the control thermostat reset,
which energized the heating element and started another heating cycle.
Another air temperature of interest is at the tumbler intake. The air temperature
in this area rose to about 175°C during normal operation [see Figure
Airflow through the exhaust duct was characterized by a constant flow,
unaffected by the accumulation of small amounts of lint in the lint screen.
The average exhaust airflow was about 1,337 ft per min (fpm). This flow
remained substantially constant throughout the drying cycle and was not
affected by the wetness of the towel load.
Pressure Drop Across the Lint Screen
When the clothes dryer was started, there was an immediate increase in
the pressure drop sensed across the lint screen. The pressure change across
the lint screen increased by about 0.025 in of water (6.2 Pa). As the load
was dried, the pressure typically increased by another 0.043 in of water
(10.6 Pa). The pressure change was not linear. Rather, the pressure difference
increased very slowly for about the first 30 min, then increased more rapidly
for the last half of the drying cycle. The turbulence of the air in this
region of the clothes dryer is evident in the signal variability [see Figure
Ambient atmospheric levels of CO2, CO,
and VOC were observed during normal drying. For CO and VOC, regular towel
loads with or without
the washing generated predictably low signals. For CO2, ambient
background levels were measured. These values varied from around 330 to
Detection of Abnormal Operation
Figure 4: Pressure Sensor Response to Abnormal Test Conditions. CLICK to
see the full-size image.
All of the abnormal operating conditions were detected by the sensor
instrumentation as a variance (either increase or decrease) from previously
nominal values, conditional on the test being executed and the sensor
being examined. Depending on the amount of “abnormality” in
the testing parameters, the sensor response magnitudes ranged from small
Tumbler Intake Temperature
The air temperature at the intake to the tumbler was a sensitive indicator
of abnormal operation. This sensor responded strongly to air leaks, exhaust-ducting
blockages, and blocked lint-screen conditions. The rapid temperature changes
during the abnormal test are caused by the cycling of the high-limit thermostat.
The tumbler intake air temperature can indicate abnormal operation in two
ways. One method is to detect air temperatures much higher than the normal
values. The other method is to monitor how frequently the high-limit thermostat
cycles. Rapid cycling of the high-limit thermostat is indicative of high
air temperatures in the area of the tumbler intake [see Figure 3].
A magnitude change in the exhaust airflow was shown to indicate some types
of abnormal operation. An increase in the exhaust airflow was a general
indication of an air leak inside the clothes dryer or a missing lint screen.
The magnitude of the change from the nominal flow of 1,337 fpm was between
+8 percent and +12 percent for the conditions tested.
Exhaust ducting blockages decreased the airflow to 1,120 fpm (-16 percent)
at a 50-percent blockage and to 625 fpm (-54 percent) at a 75-percent
blockage. Overfilled tumbler conditions did not affect the
measured airflow in this clothes dryer model. Air entered the tumbler,
traveled along the rear wall, and exited to the blower without mixing with
the towel load. Clothes dryer designs with lint screens located in the
front might show reduced airflow under overfilled tumbler conditions.
Pressure-Drop Across the Lint Screen
The differential pressure across the lint screen was an indicator of some
abnormal testing conditions. Depending on the test executed, either the
static pressure (the pressure change from off to on) or the dynamic pressure
(the pressure change from the start to the end of the test) varied from
the nominal case. The large static pressure increase associated with a
partially blocked lint screen is readily apparent. The smaller magnitude
dynamic pressure change seen with the missing lint screen is more subtle
[see Figure 4].
Sensing CO was shown to be a more sensitive indicator of combustion occurring
inside the clothes dryer than detecting either CO2 or
VOC. The relatively variable background levels of CO2 sometimes
made the detection of combustion
products difficult. In the case of CO, however, the combination of low
background levels and large magnitude changes during an abnormal test rendered
detection unambiguous [see Figure 4].
The IR sensor plus the CO sensor can be used to distinguish flaming combustion
from smoldering combustion. If a high signal is read from the IR sensor
and the CO sensor reports elevated levels, then flaming combustion might
be occurring in the tumbler. A low signal from the IR sensor combined with
detection of high levels of CO is more indicative of smoldering combustion
[see Figure 5].
Figure 5: Carbon Monoxide Detection of Smoldering Combustion. CLICK to
see the full-size image.
Sensor fusion is the combination of signals from different types of sensors
to determine an operating state. For some test conditions, combinations
of multiple different sensor readings are needed to more precisely identify
a particular abnormal operating condition. If only a general indication
of a normal/abnormal operating condition is desired, the number of sensors
required for sensor fusion purposes can be minimized.
There are various possible sensor combinations. Table 1 lists some of the
strongest-responding sensors for the selected test condition. Logical inferences
can be made between a combination of sensor responses and the functional
condition of the clothes dryer. Algorithms using the logical AND, OR, and
NOT operators may be combined with sensor readings (e.g., nominal, higher
than nominal, lower than nominal) in a truth table format to identify particular
clothes dryer operating states. This lends flexibility to product designers
in their implementations of sensor fusion to distinguish abnormal operation.
In some cases, the absence of a sensor output’s change from the nominal
value coupled with another sensor’s changed output identifies a particular
abnormal operating condition.
Figure 6: Condition-Based Monitoring of Increasing Exhaust Duct Blockage. CLICK to
see the full-size image.
The “normal” operating values of some parameters in a clothes
dryer vary from unit to unit. Aspects such as component tolerances, installation
factors, and ambient environmental conditions combine to establish particular
normal operating characteristics for each clothes dryer. Perceiving a
change from normal operation requires establishing unit-specific normal
for comparison purposes.
Condition-based monitoring is one technique that may be used to define
normal operation for an individual appliance and detect operational changes
from normality. Long-term monitoring of clothes dryer performance is useful
to identify slowly changing conditions that take an extended period to
become evident. Comparing current performance to previous operation allows
changes to be detected before clothes dryer operation affects consumer
As an example, the exhaust airflow in a clothes dryer can be considered
for condition-based monitoring. Variations in the exhaust ductwork and
the site altitude preclude the use of one normal airflow value for all
clothes dryers. After dryer installation, an initial airflow value can
be established in one of many ways. A standard type of load can be dried
to set up the initial value. Alternatively, the clothes dryer can automatically
average airflow values from the first “N” loads dried to define
the starting value. Other methods are available.
Subsequent clothes dryer use generates airflow values that are compared
to the initial measurement. Slow, long-term changes in the airflow are
easier to detect when the monitoring system retains previous readings for
In the case of exhaust airflow, detection of a decreasing airflow can be
indicative of an increasing exhaust duct blockage. Conversely, a general
increase in the exhaust airflow over time may point to an air leak inside
the clothes dryer. In either instance, condition-based monitoring methods
can be used to detect ever more abnormal operation for the monitored appliance
before conditions develop that might lead to a safety hazard. Depending
on the variables considered for condition-based monitoring, the comparison
value can be periodically updated to reflect ordinary changes during the
appliance life or preserved as was first calculated.
Data from successively blocked exhaust ducts (blue line) were fitted with
a trend line (black line) [see Figure 6]. At a fixed offset from the installed
clothes dryer’s nominal airflow, a shaded detection zone is established
in which the monitoring system detects the reduced flow and alerts the
consumer, begins to monitor other sensors to preclude a hazardous condition
from developing, or takes some other action. Once the lower operational
limit of airflow is reached (red line) for this clothes dryer, the condition-based
monitor acts to prevent further development of an incipient hazardous condition.
Table 1. CLICK to
see the full-size image.
An electric clothes dryer was instrumented with a variety of sensors and
operated under normal and abnormal conditions. Analysis of the sensor outputs
shows that abnormal operation can be distinguished from normal operation.
Multiple sensor outputs can be logically combined to uniquely identify
particular abnormal circumstances. The use of condition-based monitoring
procedures may be used to identify clothes dryer operation outside of previously
established normal ranges.
The use of sensors and data analysis techniques can potentially be used
to identify conditions that may lead to consumer hazards before the hazard
is manifested. This holds the promise of reducing the risk of fire or other
hazards in future products.
1. Mah, J., Table 1, “Estimated Residential Structure Fires Selected
Equipment 1998,” 1998 Residential Fire Loss Estimates, Directorate
for Epidemiology, U.S. Consumer Product Safety Commission, 2001.
2. Appliance.com, 2002.
3. Kadambi, S., Final Report on Electric And Gas Clothes Dryers, Directorate
for Engineering Sciences, U.S. Consumer Product Safety Commission, 2000.
4. National Fire Protection Association, The U.S. Home Product Report
(Appliances & Equipment
Involved in Fires), January 2002.
5. Butturini, Randy, Sensor Technologies to Reduce Consumer Product
Proceedings of the 54th International Appliance Technical Conference, West
Lafayette, IN, U.S., 2003.