<1>The
basic MODEL of a Greenroof is built of: hydro/thermal/bio/structural/economic
elements, each of which is a set of variables. The variables are component and
performance attributes. Component
attributes are combinations of organic and inorganic materials in the system
horizons which determine the depth and composition of soil/plant associations
and substrate components.
These are:
3 seres
(soil/plant associations); forest, meadow, seashore
4 humus; peat,
domestic compost, bark, sawdust
6 retention;
pumice, polymer, rockwool, poprocks, clay, florist foam
3 erosion;
polymer, textile, spray mulch
4 drainage;
2 waterproof;
sheet, liquid
Performance
attributes measure how these components work together in the environment of
each site to effect change in the goal variables of runoff delay, field
capacity, runoff quantity, detention time, and flow envelope per rain event.
The data will
be stored as a delimited text file (time and datavalue string per instrument),
and we will derive a quantified description of the difference between planted
and bare, flat or pitch, plant and soil types, cost to install (time &
materials / sq foot), and whatever else we want. We will develop a graphic
representation of the results.
<2A> Basic
HYPOTHESIS is that the rain on a planted roof can be held by capillary action
long enough to evaporate or be converted to biomass rather than going
downstream. This will be true up to some threshold for all rain on that depth /
composition of soil and plant associations. The threshold is quantified as
inch/hours of rainfall to reach the limit of saturation where gravity, surface
tension and friction become the controlling hydraulic mechanisms. This limit
can be quantified as the tendency for gravity (m/s2) to overcome the surface
tension (-g/cm2) and vapor pressure of moisture (%/Vol) in the
soil/plant/climatic association. This is analogous to a change in phase from
vapor (bound to air of the soil voids) to capillary film (bound to the surface
of the particle pores), to liquid (bound to itself by surface tension which
displaces air in the soil voids, and flowing due to gravity). In other words
the liquid (H2O) is interacting with a gas (vapor), a surface (film), or a
liquid (itself) as the dominant phase relation at any point in the saturation
spectrum.
All three
regimes are present at any point in time, the question being the dominant
behavior of the system. The transition
from vapor to film is a function of the dew point and void ratio of the medium,
the transition from film to liquid is also a function of ambient conditions and
surface ratio in the medium. Water precipitates from the soil according to the
same dynamics as when it precipitates from the air, being a transition from
vapor to liquid and from a distributed vector (capillary action in soil and
condensation in air as clouds) to the dominant vector of gravity as rain drops
or internal flow in the roof system.
Greenroof
systems introduce a mechanical detention into the flow path which retains water
as liquid rather than moisture held in capillary tension. The transition point
from saturation to runoff will be delayed and threshold increased by plants,
soil, and substrate components of the green roof, and the runoff above this
threshold will be lower and slower than bare roofs in similar rain conditions.
We will measure:
1) Delay from
onset of precipitation to onset of runoff for bare and planted roof systems.
The difference (offset) in time will demonstrate a benefit of the planted roof
system in the runoff impact envelope of the aggregated sources in a sewer
system. In other words, if 10 roofs runoff this fast as built, but we cause 4
of them to runoff later, what fraction of the storm event impact does that
alleviate in the watershed? This applies for delay to peak from start-of-rain
and start-of-runoff, and other measures of impact which we will monitor. (see
Table 1&2)
2) Slope from
start-of-runoff envelope to peak. The addition of soil horizons in series
between the precipitation and the bare roof system will not only delay the
transition from capillary to gravity as the dominant hydro force, but also
lowers the slope of onset which is the difference of flow per sample interval.
A planted roof system will runoff later and cause less impact than a bare roof
surface. This is the difference in Time of Concentration between greenroof and
bare.
3) Peak
amplitude of runoff compared to bare roof. The soil/bio mass will retain enough
water for enough time each rain event that even during peak of runoff, the
gallons per minute will be less than a bare roof surface.
4) Total
accumulated runoff for reference and test roof for each rain event. How much
rain fell and how much was retained by the greenroof compared to a bare roof?
This is measured as total area under the curve per rain event per roof or test
panel. We also need to measure each test roof bare, to establish a baseline of
how much water runs off how fast from the roof in a typical rain event before adding a Greenroof system. This will
also help to work out any bugs in the monitoring system before we deploy
several of them. Let's make one work on the existing budget for proof of
concept to extend to the year++ project. Then we can set them up for remote
access, and all that stuff, but first let's make them work to measure the
runoff from a bare roof on one site!
Then several sites to make sure we got the picture. Then the added mix.
<2B> Taken
with long periods of drought, this implies that our sample time density for
rainfall-to-runoff behavior needs to self adjust to its own recent history.
Since a rain gauge has a delay to first tip while waiting for rain to
accumulate, a leaf wetness sensor provides simple (cheap) redundancy to time
the onset of precipitation. No sense sampling if there's no action, and we want
to watch the time behavior of the roof system in a storm so we have to sample
every (xx?) minutes when it starts raining. We begin logging data on first tip
of the reference gauge, and continue logging (x?) hours after last reference
gauge tip. This should include all the real action from the runoff gauge, since
the rain event might go from "heavy" which will cause runoff, to
intermittent drizzle which will be retained by the planted roof system.
On a bare roof,
the runoff from heavy rain (>0.5"/hour) can fill the downspout in five
minutes. If one hundred such roofs in the neighborhood send a runoff pulse into
the sewer system, it contributes to the water shed problems that we are trying
to alleviate. If we alter (xx%) of those roofs to a detention envelope, how do
we quantify the impact on the watershed? Per cent of pervious roof area
recovered is not the only factor in a local watershed that affects runoff
impact envelope compared to the "native" ecosystem. Differences in
ground cover and soil horizon drainage which have been altered by development
are also important, but are beyond the scope of this project.
<2C> The
sample interval for each (hydro, thermal, environmental, bio, etc..) variable
will be independent of the other sample sets. In other words, we will measure
an environmental gradient X times/Hr every hour of every day to compare with
other environmental measures of (wind speed (average & peak) and direction,
insolation, air humidity, etc..) recorded at the same time. Each set of
variable strings is recorded independently and time stamped in common to allow
combinations of performance data to yield varied information (Table 1&2).
These environmental measurements, which are part of the frame of reference for
the roof system's behavior, are independent of the rainfall-to-runoff behavior,
which is only measured when it rains and is recorded in a resolution of minutes
rather than hours. Both of these are independent of biometric measures, which
are taken once per week at the very most often to show How my Garden Grows. My
Garden is not going to do any thing more spectacular than survive the first
year, and the worst case scenario of drought this first year will require
measured irrigation. This control of sample data timing will happen by software
in the site computer.
<3A> Basic
DESIGN is to fully characterize the behavior (thermal gradients, evaporation
rate, runoff amount and timing, population growth) of several sites which we
will select from the list of committed participants. Test cell...
We should focus
some effort on designing a test facility to characterize soil mixes and
moisture retention components, rather than the MUCH more cumbersome approach of
spreading the experiment all over the neighborhood. Why not build a shack in
somebody's backyard with a rain simulator and a tilt bed with full
instrumentation and control for thermal/hydro/environmental conditions that we
can apply from our knowledge of roof behavior? Instead of 20 or 30 test panels
spread out all over town, with associated instrumentation and communication
overhead, why not build an environmental test facility?
The construction
of said facility consists of PVC sprinklers hooked to a hose above a tilting
test panel (same size as the rest of them). This can be fully instrumented in
the corner of somebody's garage / shed / kitchen... and will provide calibrated
information about soil depth, mix, additives, roof pitch, and the effects of
many other variables which we want to measure. The information which we can
derive from this will be cheaper, more accurate, and quicker than from a
multitude of sites. We can even measure the exact amount of input water, which
is sketchy in situ.
We need several
test panel sites for various reasons (including bio questions, installation
costs and techniques, public photo ops, etc) but offloading component tests
leaves more budget for performance variables in real world conditions. One of
the major limits which has been discussed in planning is lack of funds for soil
and plants. This added approach reduces the need for plants and soils in
quantity, and allows some of that budget to be shifted to allow more different
types of soil, substrate, and sponge to be measured together. We still need to
see the roofs and sheds and panels, just not so many of them to get more better
information quicker at less cost. In one afternoon, we can measure the exact
runoff response of a panel configuration and then measure evaporation overnight
inside the shower curtain. With greater accuracy and control in our method, we
need fewer data points to achieve the same certainty of information. That means
we have a benchmark range for each test site to demonstrate in real life,
instead of using the sites to gather the basic data for measurement.
We will also monitor just the runoff behavior at three
additional roof sites, and we will use the rest of the sites and test panels to
develop data for the Architectural, Biological, Economic and Agronomic
measures.
There are 11
environmental variables that we can measure to characterize the behavior of the
system: reference rainfall, runoff flow, soil humidity, air humidity, air temp,
soil temp, interior temp, wind speed, wind direction, insolation, and
barometric pressure. By measuring these relative to each other as a time
series, we will eventually build enough performance data to have statistically
valid information that can be compared between sites and component sets. See
Tables 1&2 below.
If we measure
with the same setup and procedure, then we can compensate for the effect of
differences between sites which leaves us with performance under similar
conditions. Component variables have to be constant for at least two of the
fulltest sites for comparison over time to validate the measurements. One of
the fulltest sites with similar soil composition and depth should ((include
commercial components)) to compare published specifications and current market
practice with our experimental generic components.
<4A> HYDRO
variables which we will focus on are the comparative runoff envelopes in a
series of rain events for the planted roof compared to an equivalent bare roof
at the same location. The equivalent bare roof can be extrapolated from a
calibrated sample area (per cent of planted roof) into a rain gauge. We want to
establish a data base which compares runoff retention and detention (how much
and how long) for different combinations of soil depth and composition, and
different plant associations for the given exposure and climatic conditions. We
are measuring the total area under the curve of runoff per event, slope of the
curve from start of runoff to peak runoff, delay to onset and peak of runoff,
and difference in the peak runoff amplitude between bare and planted. Complete
characterization requires that we measure the behavior of the bare roof before
installation in a series of typical rain events for the local climate in order
to have a benchmark for that particular site and roof structure. Without this
frame of reference, measurements exist in a vacuum.
With the information in place, we can extrapolate how
any roof in any rain event up to 2 or 5 year storms will behave in usual
circumstances with a particular combination of components. This climatic data
is available on the web for many locations (monitoring stations) in the western
USofA at: http://www.wrcc.sage.dri.edu/summary/climsmwa.html
The information contained on this site tells mean/min/max precipitation for
last 25 years, which is a frame of reference for what the roof system will
encounter. In which months will the typical rain per day exceed
.01"/0.1"/1.0" etc.? Max rainfall in Seattle Urban measurement
zone for the last 25 years happened on Jan 18, 1986 with 4.22" in one day.
Good surfing on Broadway. How to configure the roof components to handle an
event of this magnitude? Probably not, but the typical rain event is much less
than this and a realistic composition can be targeted to handle most rain
events up to some max, with a reasonable level of confidence. Without data, we
can only guess what scale this cycle of
precipitation->retention->runoff->evaporation happens over what period
of time? When we have “enough” measurement data, we will know whether it’s
liters per minute or milliliters per hour of system performance, and just how
different are the components from each other? This provides a design goal
within which the component variables are tested and adjusted to establish
performance of a particular combination.
<4B> We
will measure the delay between onset of precipitation and onset of the
transition from capillarity to gravity as the dominant hydraulic mechanism. We
also want to measure the other delayed transition point from capillarity to
humidity as evaporation removes water from the soil until it reaches
equilibrium with air relative humidity. For this purpose, these two points are
defined as the transition from (xx%) capillary saturation to onset of runoff,
and transition to <100% relative humidity of the soil. This describes the
behavior of the system when gravity (runoff) and evaporation are the forces
which control the rainwater in the soil mix, and provides a reference baseline
of inorganic soil hydrodynamics to define differences introduced by plants and
other organic components. The third transition is from capillary moisture to
the "wilt" state, where there is insufficient water to maintain cell
structure for a long enough time to permanently affect plant growth. At this
point, the soil is in negative vapor tension to the plant cells, and will draw
moisture from the plant system.
<4C>
Transpiration by the plants is so much slower and lower than gravity and
evaporation as a transport or retaining mechanism in the roof system, that we
will decline to include it in our calculations. We assume, for this study, that
the major contribution of plants in reducing runoff is mechanical in action by
increasing internal surface area (root mass and leaf/stem area as void ratio)
to the roof system. This is compared to the much smaller and slower bio-process
of osmotic transpiration through the plants. The other mechanical contributions
of plants are in holding the soil together to prevent erosion by wind and
water, and in the environmental barrier of biomass which affects evaporation by
limiting exposure to wind and sun on the soil surface.
<4D> We
can also measure humidity in the soil as a function of time,insolation,
temperature, wind speed and direction (exposure), and air humidity to determine
the rate of evaporation for a given soil and plant association. Live and dead
plants on the soil surface contribute more to maintaining water in the soil by
slowing down evaporation than they convey out by transpiration. This happens
because the plants and mulch shield the surface from exposure to sun and wind,
which keeps the temperature of the soil lower and prevents wind conveying
moisture away. Measurement of this performance variable in the first year when
the plants won't contribute much shade or mulch gives us a baseline to compare
future behavior after the bio components become established.
<4E> We
will test the idea that the influence of roof pitch is measurable only in the
runoff envelope. The capillary and evaporative mechanisms are only slightly
influenced by gravity. This means that the same soil composition and plant
association in the same conditions will behave the same up to the limit of
capillary saturation, at which point runoff begins regardless of pitch. This
will appear as the same(?) amount of runoff, delayed the same(?) time from
start of rainfall, but with a higher peak and shorter timebase associated with
greater pitch (more vertical). If the roof surface is normal to the prevailing
wet weather wind direction, at what wind speed does the incident precipitation
increase when roof pitch presents a larger cross sectional area to the weather
system? Is there a measurable difference of incident rain falling on a flat
roof, a 4:12 pitched parallel to prevailing weather, and 4:12 pitched normal to
prevailing weather? At a sustained wind
speed of 20 miles per hour, rain drops will deviate from vertical by some
angle; if the roof is pitched to a ratio of 4:12, how much more rain will fall
on that surface than a flat roof? Same amount? How about 1:1 or 6:12?
<4F> For
pitch, we will measure three test panels of the same size and composition,
parallel to ground, 6:12 pitch and 1:1 pitch. Instrumentation is a tube turbine
with a Hall effect counter, at the runoff collection corner of each test panel,
which totalizes revolution pulses per period into a computer (see Omega
Flowmeter spec). Download each week or each month. Total runoff is logged by
pulse per minute, we are only interested in comparing three panels to each
other and to the results of the same composition in the test cell.
<4G> Total
cost of test with construction costs included is <$200 per test panel to
have some reliable information about the influence of a simple variable for a
given soil/plant mix. It would be best to replicate the set up with one
component changed the same amount on all three of the comparison panels at a
site. In other words, does the same behavioral relation happen among the panels
with a different inorganic detention additive in the mix? If the A horizon is a
consistent soil mix, with a typical planting density which won't change for the
first year, then we can compare different components of the B or C horizon to
measure relative effect on the hydro variables.
<5>
THERMAL variables describe thermal conductivity and capacity of the organic and
inorganic components compared to a bare roof. We can measure temperature
simultaneously (or close enough) at a short distance above the bio mass,
between the bio mass and the substrate, and a short distance into the interior
space. Difference between these temps at the same time gives an indication of
the transfer function of a greenroof compared to a reference bare membrane (or
composition) roof structure. The challenge in a reference test panel for
thermal transfer measurement would be to isolate the test surface from heat
loss through the other sides of the box.
In order to relate the measurements, we need to know the thermal
conductivity, diffusivity and mechanical structure of the soil. Temp difference
across 10-25cm of consistent soil medium is likely to be in the single digit
range, it's not as important as the difference in rate of heat loss or gain
effected by the greenroof. We will quantify thermal behavior of various soils
and plant associations in combination with drainage layers acting as a system
of conductive and convective transfers. When the drainage layer is dry it acts
as a dead air space insulator, and when it is draining it acts as a coolant
flow conveying heat from the system. Might be a good reason to put insulation
underneath?
<6> BIO
measures the planting density and propagation rate for a specified association
to reach full layered density in the niches: Moss, Succulent/Sedum, Herbal,
Vine, Grass, etc.. Also, it's necessary to characterize root layers, soil
composition, pH, organic and inorganic nutrient structure for plant
associations. It's an easy addition to characterize plants for bird and insect
habitat, but that's another budget. We need to arrive at a design for the soil
which optimizes for cost, weight, water capacity, and nutrients. It is CRITICAL
to describe the medium in which hydro action happens!! That includes the
biomass horizon, which extends above the soil, and the drainage layer below the
root barrier.
<7>
STRUCTURAL components will include static and dynamic loads, physical layers,
placement of loads relative to the existing structure, and more architectural /
engineering stuff (a little help here..?) Drainage and root/sedimentation
barrier which is the equivalent B/C horizon, has several available components
both proprietary and generic. We should compare the cost/benefit of different
filtration and drainage technologies to provide the consumer a means to make
decisions based on their particular goals and budget compared to what is on the
market today.
<8> We
need to know what resolution of (rainfall, temp gradient, Lbs/sq.ft.,etc.) per
what unit time will give a useful measure to compare applications with each
other and with a reference. How do we know we're measuring with the same
yardstick? In other words, for a given roof pitch and footprint what variables
will optimize the hydro and thermal retention effected by the bio layer
compared to a similar bare roof? In what time frame for how many dollars and
labor hours per square foot? How do the load vectors add to a given (as built)
roof? Any reasonable structural improvements available to support a greenroof
vs. standard construction? Could it be worth a roof remodel to have a garden up
there which has energy/environment/economic benefit to the homeowner?
<9> We
will start with an extension of measurement parameters (in scale only) from the
project report "Field Eval Perm Pave Sys.." (Booth,Leavitt/JAPA V.65
#3), which used a dual tipping bucket gauge of 0.1 liter tip on a 15 minute
time scale to develop a database to characterize difference between previously paved versus pervious.This
extension, which was developed to compare rain event runoff envelope for
perv/imperv areas in a parking lot, is valid in data scale and structure. Scale
is the question of match between event and measure, i.e. quantity / time or
"How many mL/15 minutes (data logger period) into the measure, compared to
how many mL/15 minutes rain through the system?" Structure is valid
because we're comparing reference runoff to treated equivalent area. They did
1:1:1 of 2 different treatments to 1 reference equivalent area. It may be
useful to reference the local TV weather radar history?
>>TEST
PANELS will be grouped by Soil, Plant
Association, and Pitch.
>Soil has two
component sets, inorganic and organic, and two performance sets, detention
value and field capacity. The A horizon is the organic component of the system,
with inorganics added for structure and water detention. This is where most of
the biotic action happens, and provides much of the surface area for capillary
water retention. The organic components are mostly fibrous
plant material which acts to raise the Plastic threshold
of the soil by giving humus and the inorganic aggregates something to cling to,
like organic rebar on a microscopic scale.
<10A> The
inorganic components will contribute two controls on the roof system:
1) porosity to
retain max water in capillary distribution per unit volume of the system, each
type of component has a different uptake rate, capacity, and dry out rate
(hygroscopic transfer). These components are in five categories on the market:
Polymer Gel, Mineral Wool, Clay based (Zeolite, Sintered pellet, Cat litter),
Perlite/Vermiculite, and Florist Foam. We will measure comparative behavior of
Polymer Gel (Soilmoist), Mineral Wool (x?), and Perlite. The difference between
Perlite and Vermiculite is primarily in the hydro dynamics of each. Porosity
and the time behavior of water transfer is completely different, and
potentially significant in a greenroof application. But we're not going to
measure that here on this budget, because they already did.
2) Drainage
texture to retain runoff and nutrients, while it protects the base membrane. If
the organic components of the soil are the same for different inorganic
components, then the comparative detention behavior of each inorganic component
will have plenty of time to be described before any plant association has time
to mature into significance.
<10B> We
will monitor a max of three types of inorganic component (polymer,
Perlite/Vermiculite, and mineral wool) for their relative detention value under
similar conditions, and can use the same three panels to measure growth rate
and survival rate for three different plant associations which are appropriate
for the common organic soil mix. At least one of the inorganic components
should be a commercially available greenroof system, with the common organic
mix, so that we can extrapolate generic component behavior (somewhat) according
to their product test data.
<11> The
difference between reference total precipitation and runoff total is the added
water load to the roof at 8.3 Lbs/Gal (62.43 Lbs/CuFt). Differences between dry
bulk densities (weight/volume) of inorganic components which expand the field
capacity of the system, will not be large compared to the relative added load
of water by unit volume which tells us the relative retention for each
component. In other words, per cubic foot of the greenroof system the difference
of half ounce dry weight between component A or B, is less important than a
half gallon of water retained by A more than B. Most cost effective inorganic
components will add approximately the same static dry load to the roof
structure, if applied in the same proportions to the soil mix. We will measure
the difference in dynamic load according to the difference in time and per cent
threshold of the transitions in and out of capillary saturation, by
transportation and evaporation but not by transpiration. These three mechanisms
account for all the precipitation and irrigation water input to the roof
system.
<Table 1>
MONITORING VARIABLES:
VARIABLE
MEASURED SAMPLE RATE
ATTRIBUTE DATA
UNIT
A) INSOLATION W/m2 1
hour
B1) TEMPinside
Centigrade 1 hour
B2) TEMPsoil
Centigrade 1 hour
B3) TEMPsurface
Centigrade 1 hour
B4) TEMPoutside
Centigrade 1 hour
C1) HUMIDITYair % RH 1 hour
C2) HUMIDITYsoil
% RH 1 hour
D1) WINDspeed m/Sec 1/2 hour
D2) WINDazimuth Polar 1/2 hour
E1) RAINfall mm/Hr 1/10 hour
E2) RAINrunoff L/Min Minute (?)
F) BAROMETER cm
Hg
6 hours
<Table 2>
INSTRUMENT/VARIABLE: (all are measured on same timebase in monitoring computer)
VARIABLE INSTRUMENTED MEASURED
MEASURED ATTRIBUTE<Table 1> UNITS
Time to Runoff Start [E1,E2]
Min
Time to Runoff Peak [E1,E2]
Min
Slope max to Runoff Peak [E1,E2]
L/Min/Min
Height max of Runoff Peak [E1,E2] L/Min
Total Event Runoff [E2]
L
Total Event Rainfall [E1]
L/m2/Hr
Evaporation Rate [A,B2,B3,C1,C2,D1,D2,F]
%RH/Hr
Pitch Exposure [D1,D2,E1,E2,F] further definition
Thermal Transfer [B1,B2,B3,C1,C2,D1]
further definition
<Table 3>
SENSORS:
MEASURE MODEL COST
FORMAT WWW page per site??
A)
INSOLATION analog?
B1)
TEMPERATUREsoil National
Semiconductor LM35 $10 analog
B2)
TEMPERATUREair Oregon
Scientific + LM35
C1)
HUMIDITYair Oregon
Scientific
C2)
HUMIDITYsoil Hygrometrix HMX
2000 analog
D1)
WINDspeed Oregon
Scientific pulse
D2)
WINDazimuth Oregon
Scientific analog
E1)
RAINfall Oregon
Scientific pulse
E2)
RAINrunoff Omega FTB
2005 $75 pulse
F)
BAROMETER Oregon
Scientific
The cost per
site for measuring runoff behavior has two components, the number of sensors
and how the software is distributed. We can spec intelligent sensors which will
automatically count/total/format and transmit the pulses as RS 232 information,
or we can use dumb sensors which only transmit a pulse and leave the software
overhead to the computer. Self contained dataloggers are expen$ive little
things as are intelligent sensors. Also, a nicely packaged consumer weather
station readout panel which feeds proprietary data into a Windows application
is available from several manufacturers, but the minor convenience of separate desktop readout is not worth the
loss of scalability and design in the monitor system. The better option is to
use dumb sensors, and put as much of the software processing into the computer
as we can. We can use the serial and parallel ports for input, allocating
sensors according to their output format: analog converted to digital data
packets into the serial port, and pulse or state changes into the parallel
port.
Fortunately,
the measurement of precipitation and runoff is a pulse train from each sensor.
Unfortunately, the pulses are completely independent of each other in time
(phase/frequency/start/stop), which means we need to be relatively fast in
scanning several of them to make sure we get the pulse data from all of them.
We can use the parallel port as an input for 8 pulse sources and 4 state
sources. Full spec on this mode is in IEEE 1284 Level 1 Byte Mode. This approach
requires a program in DOS or BIOS rather than Windows, because DOS/BIOS is
simpler / faster / less crash prone / sufficient. Once the data is in the
computer with time stamp, we can do the manipulations to make it information.
This saves the communication overhead, and gives us all the info we need for
runoff envelopes. The hardware is three noncritical wires to each sensor
(power/data/ground), the parallel port connector (D 25), and a power supply
buffer (7805 voltage regulator with a couple of resistors). Simple, reliable,
cheap. Invest in software instead of hardware.
We could use
the Control Register pins as power sources for the pulse or state sensors, but
limits on output power and protection for the computer from environmental
mishaps are too possible for us to do that here. Instead, we use an available
ribbon from the power supply with separate output isolation to drive the sensor
array. This gives us power to the pulses with system signal ground and no
outside interference, and it's included free with our hardware package
purchase. This is a good thing technically and financially, because it solves
the problem of getting the data into the computer in a format that can be
readily processed into information about environmental events of interest.
The analog sensors (Temp / %RH / Solar / Wind
Azimuth / etc..) can be connected to the Dataq DI-194 which has onboard A/D
conversions with 8 bit resolution for 4 input channels at 240 samples per
second, and RS 232 output of the converted data value. All the analog values we
want are slow changing, so speed to read is not as much of a challenge as the
pulses. The DI-194 can be networked (RS485) if we need to expand, and they only
cost about $20 each. They will input the data through the serial port as RS232
packets.
This is by far
the most cost effective approach, and it is scalable and easy to replicate
(<100 lines of DOS code) for all the sites including the reference cell or
student volunteers. More to come..
Also, a chip
(National Semiconductor LM 35) converts temp to an analog value, is cheap, and
requires very few external components to be used. The benefit is that we can
measure several points of temperature on the roofs (not panels) to look at
thermal transfer rate for less money than it would cost to use one
"weather" thermometer. I can do the circuit and fabricate the unit,
and they don't require calibration or maintenance.
The computer's
perspective in all this is like watching a movie in a room with several lamps
which turn on or off sometimes. The movie consists of eight little windows with
strobe lights behind them that flash sometimes. Good wholesome entertainment
for a computer (or an accountant). When a lamp turns on or off, our little Guy In There (GIT) makes a mark
in the appropriate column of a form and likewise for the windows flashing.
Occasionally, some other git comes in and takes the form away. When this
happens, our movie watching git gets to work on a clean form. The used form is
taken to the secretarial pool where we have one secretary dedicated to each
lamp and window. These lucky secretarial gits count the number of marks in
their column and send that result to the accounting department and management.
The managers are interested in whether the lamps have turned on or off and what
time is it, while the accountants only care how many flashes and when the
number came through. All the rest is storyline, as they say, and happens in the
statistics which we apply to this data.
Comparing this
to a movie is important because a movie happens at 30 frames per second, which
fools our happy little brains into thinking that there's something moving on
the screen. The analog is that the software port driver cycle time is
equivalent to the frame rate of the camera which must be fast enough to
reliably catch the flash in eight windows at random times. We don't know when a
flash will happen but we know how long each flash will last so we need a camera
which will take at least 2 frames per flash. This means we need a short
executable loop in software which reads the parallel port input register fast
enough to capture each incoming pulse twice (up edge / down edge), and compare
to the previous read for a reliable decision as to whether a pulse has
occurred. That result is what our git
marks in the column, or not. The limit on git's performance is that it
takes a couple of instructions to make this all important decision as to pulse
/ no pulse, which takes time and attention from marking it down or watching the
screen.
We can provide
some help in added hardware or code (8 window watching gits, who tell the
logger git when they know a flash happened) but there is always an upper limit
to how much / how fast. For our purposes, this will work if it happens in DOS /
BIOS plus we can do a little number crunching for more efficient storage and
processing of data into information.
The lamps which
go on or off are state sensors, which tell the managers that something is
different from the last time we checked. "The water level in the runoff
cistern is too low to pump, so don't even think about it. But the sun has not
shined for three days now, so don't worry." Maybe we could set an alarm
condition which would result in the computer executing 10,000 iterations of a
rain prayer, like a cyber-prayerwheel. Probably, we shouldn't. These lamps are
the Status Register, which normally tells the computer when a peripheral is
ready for more data or is out of paper. The Leaf Wetness sensor is one example
of a state change lamp, which sets a flag for the computer to wake up and watch
for a Tipping Bucket event, which tells the computer to
watch for Flow pulses until the Bucket no longer tips and the Leaf is dry. This
saves memory space from being filled with non-event data (zero pulses), which
is useless overhead.
The flashes of
light in the 8 windows are of course any sensor which puts out a pulse as the
result of an environmental event, like another hundredth of an inch of rain or
a silly milliliter of runoff or wind speed, etc. These are the Data Register,
which makes the computer think it's seeing data being transferred in, rather
than a bunch of analog hardware pulses which is what's really happening. The
computer transfers the data as IEEE 1284 eight bit (plus the
Status register flags) input,
The analog data
which we gather from moisture sensors or wind azimuth etc. comes in on the
serial port as pre formatted eight bit data from the Dataq DI194. This unit is
also powered from the serial port, and comes with software (Windaq Lite) for
240 samples per second of signals on 4 independent channels ranging between +/-
10 volts with eight bit conversion accuracy.
There is third-party software which extends the capability of the
software which comes free with the package, but we need Windows 95+ installed
on the system for any of the software to run. The benefit of this requirement
is that the data can be imported directly into Excel which makes processing,
storage and communications a lot easier. Prices for all these things are in the
$20 range as advertised, but I will talk to the various Sales Departments about
a price break when I have the design locked up, so I don't look like an idiot
when I tell them we're an Educational 501(c)(3).