|
|
_______ Q. A. _______ Q. A. _______ Q. A. _______ Q. 1)
Can you create your own materials in the library with custom properties? A. 2. There
are really no physical scale limitations (lengths could be e.g.between 1E-5 and
1E5 meters). 3. See www.buildingphysics.com
for info. 4. In
HEAT2, boundary conditions, heat sources and internal areas of a specified
temperature may be a function in time (sinus, step-wise constant or step-wise
linear). The function editor (text file with step-wise constant or step-wise
linear values) can import different formats, data may be cut and pasted from e.g
Excel. See the manuals for more info. 5. A
surface resistance coefficient (inverted heat transfer coefficient) is given as
a fix value. _______ Q. A. _______ Q. A. Q. 1) In my
typical application I have a BGA IC (ball grid array integrated circuit) or flip
chip, in electronic application, where in X axis I may
have 13 solder balls (could be 20 as well, same in Y direction), each
ball (small sphere) being
represented for instance by 4 cells in X (sphere
represented by box), and they are spaced by eg 4 cells also. If I
understand correctly, for 20 balls I will need 40 segments in X at
least (one segment at least for each material transition, right ?).
There may be 50 to 200 such solder balls in my model in the XY plane.
Defining the geometry for repetitive structures should be easy since I
can write a small C program
to generate the DAT file. My PC has 2 GB RAM so from what you said I expect that
the RAM only will limit the cell number to about 400 in X,Y,Z. 2)
Boundary conditions in Heat3 can be defined with functions, but only function of time it seems, I would have been interested in BC where the
thermal resistance toward ambient (or flux) is a function of temperature, such
as: Q in Watt = dS x Constant x (
(T+273.15)^4 - Constant) +
dS x Constant
x T^0.25 (where dS is cell area). Steady-state only. Did you receive any
request from other customers for such a feature, and would it be hard to
implement ? 3) I may
use dimensions as small as some microns (silicon chip...). Is
there some round-off truncation in the software which could lead to
errors for small dimensions like these ? A. The
amount of RAM is roughly N*32+10 MB. Note that the amount of cells does not have
to be equal in all directions. E.g. Nx=500, Ny=500, Nz=50 would only require
about 410 MB RAM. Feel free to specify your own data fields. The text
editor can hold 16 MB of text (160 characters per row would give a maximum
number of rows of 100 000) so this should present no problem for you. 2. It
would be possible to implement a heat flux according to you formula. No other
customer has requested this kind of boundary condition. 3. I
think it will be ok, but I cannot answer this question since problems may arise
with a large amount of cells. See answer 3. _______ Q. Rather
than fumble about, I decided to run your SLAB file with a mean temperature of
9.4, a period of 1 year, amplitude of 10, and phase of 0, after an initial
steady state run without modification. I changed only the one boundary
condition. Running for 59m19s of CPU time, ICPS=185425, I managed to simulate
only 1m22s of virtual time. That means (unless the relationship between real and
virtual time is non-linear) that temperatures after 10 years would take only 600
years or so to predict! My own models, even when greatly simplified, were taking
similar amounts of time. Is this reasonable, or should I be looking for
something wrong with my equipment? I am running under Windows 2000 Professional
on an AMD Athlon 950 machine. If my
equipment is performing to expectation, I must abandon the idea of doing 3D
transient calculations to compare temperatures at external corners with and
without insulation, unless you can see a way. With very low thermal capacities,
the calculations are within the realm of the possible, but I assume the results
are spurious. Can you confirm this? I'm not interested in flow as much as
temperature. My next
idea is to use 2D transient calculations first, and then find an exterior
temperature that produces the annual minimum footing temperature midway between
corners with a 3D steady state calculation. Would it be reasonable to assume
that conditions at the corner predicted by the same means would also approximate
the annual minimum? As
always, any suggestions will be most welcome. A. My
machine is a PIII 500 MHz. Your PC should be almost twice as fast (I guess about
1.9 times). I am a bit puzzled that you get an ICPS (Iterated cells per second)
of only 185425. It should be about 1.2 million if we look at the clock rate for
the PIII and Athlon (I have made some test on a Athlon 850 which was 1.8 times
faster then my PIII 500 for some HEAT2 calculations). Maybe you have other
resident programs (Norton utilities?) that takes up CPU-time. What
time does it take to solve the enclosed HEAT2 problem "bench.dat" on
your PC? Just load the file and press F9. It takes 29 sec on my PC, 2678
iterations, N=25600 cells, HEAT2 ICPS=2.36 million. There are some additional
benchmarks in the manuals. If you
are interested in temperatures close to the ground at the corners (maybe within
1 meters from the ground surface or so) I think that true 3D calculations are
best. There might exist some 2D tricks in certain case, but I am not familiar
with those. _______ Q. Using
Heat3, I am going to try looking into the effects of a combination of insulation
under slab on grade, with perimeter insulation on a horizontal plane in the
ground extending beyond the building, including a building corner. (Given the
speed of my computer, and my age, I wonder what is the probability of completing
this task in my lifetime?) To speed
the process up, I've started with a 2 dimensional model, steady state, with the
exterior temperature at annual average, to get a starting temperature field. I
am now doing a transient run, with f1=0 (i.e. starting with August, more or
less), to stop at a month. I am starting with August in hopes of minimizing
differences between steady state and transient temperature fields. It is clear
from the initial progress that to run a full month will take a long time!
(So far, more than an hour to do 10s of model time.) I can speed this up
somewhat by removing device drivers and background processes, and refraining
from doing other tasks, but this seems a faint hope. Can you suggest ways to get
a less accurate answer in less time? The full program I am trying to follow is: 1. What
minimum temperature does the footing bearing surface experience during the
annual cycle, using a 2D transient model? (At the moment, I'm doing this with
Heat3 and a 1 cell vertical slice from the 3D model with all of the sub-grade
insulation having the same properties as the ground, and the number of cells
reduced. 2. With
a full 3D model, under the same conditions, what is the temperature under the
footing at the corner, and how far back along the wall does the influence of the
corner extend? 3. What
steady state exterior temperature will produce the same temperature at the base
of the footing as in step 1, in 2D? 4. In 3D
steady state, will the same exterior temperature produce the same corner
temperature field as in step 2? 5. In 2D
steady state, if insulation is added under the slab, what additional insulation
is needed outside the building to restore the footing to the same temperature as
in 3? 6. In 3D
steady state, with insulation under the slab, what augmentation at the corner is
needed to restore it to the same temperature as in step 2? Any
suggestions you'd like to share will be welcome. A. When I
have problems of this kind with frost heave, I usually start from the
steady-state temperature (with the average outdoor mean value) and then continue
with transient calculations with a sinusoidal maybe 3-5 years until more or less
the same variation is reached. After that I apply a sudden cold-spell (e.g. -18
degC) for 1 or 2 weeks and look at the isotherms under or near the house. A rule
of “Swedish thumb” is that the zero degree isotherm must not reach
below/under a 45 degree line that goes under and out from the side of the house.
However if the soil is not sensible for frost heave the zero isotherm may go
beneath the house. The corner (3D) is normally used when looking at worst-case
scenario. The second part of the German report you got deals with added
horizontal insulation outside the corner of a house. Even
if transient calculations take time, I think it is better to do this than to try
to approximate the temperature field close to the slab with steady-state
calculations. _______ Q.
Continued from last topic... I
started on the same path anyway, although I failed on the first attempt to
change the boundary condition to "function", an error I'm not likely
to repeat. A soils consultant has suggested soil with conductivity of 1.4 W/mK
and capacity of 2.35 MJ/m3K to represent the local clay till. The current state
of the art in rules of thumb here is that next to a heated building, a
foundation depth of 1.2 m below grade is adequate. There is a similar rule for
isolated foundations, but the depth eludes me at the moment. Since, if the soil
is moist, the advance of the 0 deg C isotherm (and it's subsequent retreat)
would be much less rapid than the transient calculation will indicate (due to
heat of fusion). I still intend to assume that even if freezing is indicated
within the splay of the foundation by the transient calculation, that things are
OK, since these foundations perform adequately in fact. Moisture (increasing
capacity), freezing, insulation due to snow, solar heat, and surface resistance
are all factors that are not fully accounted for ,and that tend to make the
transient calculation conservative. Thank
you for reminding me of the bearing splay of foundations - I know this but had
not related it to this context. Unfortunately, no one has volunteered to
translate the German paper for me. I
understand and appreciate your caution, although I may not follow your sage
advice in this regard. Even with a much faster computer, I wonder how long it
might take? A. HEAT2
and HEAT3 optimize performance for Pentium II and III (and probably the same for
P4). A Pentium III 500 will probably be 4-5 times faster than your Pentium 233.
See chapter 7 in the HEAT2 manual for more info. The
clock rate is the most important factor for PII and III systems, i.e a PIII
running on 1 GHz is twice as fast as a 500 MHz CPU for calculations with HEAT2
and HEAT3. Adding more memory will not increase sepeed. 64 MB of RAM is
sufficient. It
should be sufficient to run the 3D frost heave problem for a simulation period
of 3-5 years overnight. You will have to elaborate on the numerical mesh and not
choose a too dense grid. You might end up with a 5% (or even 10%) numerical
error but this should suffice. After all, the uncertainty of the soil properties
are probably much higher than this (in addition you have all the conservative
factors you mention below). You can estimate the numerical error by doing two
fast steady-state calculations, one with the mesh you intend to use in the
transient calculations and one with a denser grid that gives a "true"
value for heat flows/temperatures. You then have an indication of your numerical
error for the used mesh in the transient calculation. I think that this error
will give reults on the safe side; a sudden cold spell will give a larger
penetration depth using a less dense grid compare with a more dense grid. But
since the outdoor temperature varies a few years sinusoidally, this may not be
true. Let me know if you find out anything about this. _______ Q: I have
made the enclosed input. The input would satisfy the HEAT3-50 conditions. When steady state solving on HEAT3-50 or HEAT3-c, I receive an error message: Too
many internal cells: (2326) Maximum = 1000. In the manual no restrictions
are made for a number of internal cells. The expression 'internal cells' is not
defined, so I do not know how to alter this number. Please
help us immediately to solve this problem. For your immediate collaboration we
thank you very much. A. "..
Seeking cells at internal corners... Found=57 OK" In the
future you may also want versions with more cells. The limit is the PC memory
size (RAM). The following list shows some examples for HEAT3:
Note
that the amount of cells in the three directions does not have to be equal. _______ Q. A. _______ Q. A. Q. A. There
are often different standards in different countries and you should normally
follow the one adopted by Germany. Here is a list of some related ISO standards.
HEAT2
5.0 has also possibilities to calculate heat transfer within frame cavities
according to the current proposed standard prCEN/ISO/TC 89/WG7, document
prEN10077-2 ( (Thermal performance of windows, doors and shutters - Calculation
of thermal transmittance - Part 2: Numerical method for frames). _______ Q. A. _______ Q. A. _______ Q. If the
drawings changed for version 10077-2:2000, please send us the result for total
heat flux (W/m) and U-value of the frame, as mentioned in the new version of the
standard (which I don't have). If the
drawings didn't change, is it possible to send us the intermediate and final
results for our dxf-attachment? Thanks for helping us in the evaluation of your
calculation program. A. Q. When I
prepare bitmap images to compare to each other, I need to get back to the same
viewpoint, zoom, scale, isotherm, rotation, etc. settings as previous sessions
to create comparable images for conditions I did not think of before. Every time
I close a file, these settings are lost. They cannot be retrieved either, in the
case of click and drag settings, with adequate accuracy. I need to open a DAT
file, load a PSE file, and then ....? to get the same image I had in a previous
session. Even the ability to open a DAT, and then load PSE files one after
another without resetting the graphics would be very helpful. Please tell me I'm
being stupid again, and how to do this! 2. When
images are saved as BMPs, only MS Paint is able to open them. Graphic Workshop,
Adobe Photoshop, MS Powerpoint, and other programs I've tried to use them with
(and which normally open BMP files without problems) all report that they cannot
be parsed, are damaged, or they open the file and display garbage. The number of
colors is set to 254 (why won't Heat accept 256?) and I believe some of the
other programs can handle up to 24-bit depth files. Any idea what's the problem? A. 2. I could
however not import it into an older version of Adobe Photoshop (ver 5.5). There
are two simple solutions: 1. Cut the image to the clipboard (see the HEAT2
manual, Section 5.18.3) or 2. open the bmp-files in MS Paint and save them as jpg
or tif that can be imported into other programs. Regarding
the colors; maybe you are using 256 colors in Windows? It should be set to at
least 65 000 ("Hi color (16-bit)"). Actually, the 254 value denotes
the number of shades in HEAT2 that are chosen from these 65 000 (or more)
colors. It would make any sense to use more colors since it is hard to see the
difference between the shades (at least with the false color representation I
use). If you set the number of colors to 100 in the scale option menu, you will
hardly see the difference. Q. A. Column 2. Show graphics for this column. Copy this to stack with the clone commando. Change color to green. Show data for column 3. This will show the
picture below.
|
|
