John R Wilson's Blog

Insights into the practical application of CFD to the thermal and airflow design process.

9 August, 2011
Welcome to Part V of a multi-part series, where in this installment Travis takes a look at his lawn mower.

One of my first jobs as a new home owner was to mow the lawn.  As it happens, we had a friend with a lawn mower that was no longer using it, so he gave it to us.  When I went to pick it up, he was using words like, 1992, and “never maintained”, so I knew it wasn’t going to be a super awesome lawn mower.  Still, I love to get something for free, and a lawn mower is a couple hundred dollars, so I was very happy with his gift … until I used it for the first time.  Granted, with all the renovations at the house and moving in, the lawn was overdue for a cut.  Plus it was just after the rainy season here in the bay area, so the grass was thick, long and wet.  How long?  In some spots it was taller than the lawn mower wheels. The issue I had with this 1992 lawn mower was it was a mulcher.  I’ve never used this type of lawn mower before, but the idea sounded great.  You cut the grass, and the clippings are put back to the lawn so you don’t lose the nutrients. Sometimes an idea is better in theory though.  It’s not that I had to pull start this mower every couple feet that left me with such a bad opinion of my new used lawn mower.  It’s all the dead grass that resulted.  You see, it seemed to get clogged up with the wet grass, and once it started to stall, it would cough up a bunch of wet grass phlegm clumps all over the place.  On the sidewalk, driveway, my car, … but mostly back into the lawn.  The thing was, these clumps were wet and heavy and big, so they sunk to the bottom of the lawn, and even after raking, many were missed causing dead patches all over my lawn. I decided to do some analysis on this lawn mower, to see if there was a design flaw with this mulching type of lawn mower.  Luckily, I found an already built, free 3D CAD file of a generic mulching lawn mower on one of the CAD user community websites (  I used our FloEFD software, which is our general purpose, CAD embedded CFD tool.  I had the CAD geometry so my time commitment to this analysis was greatly reduced.  Any designer wanting to do an analysis in FloEFD would be in similar shoes, having 3D CAD model and wanting to use that model without having to do any importing/geometry clean up or creating a “fluid” part.  All these steps are avoided when using FloEFD.

FloEFD Lawn mower Model

 As I’m not a lawn mower designer, I’m not familiar with all the details that are required for an analysis, so I had to do some research on typical RPM (~4000), and estimate the airflow resistance for the grass.  Once I had analyzed the airflow, I seeded the solution with particles of grass to see where they went.  I used fairly big particles for the grass, as my grass was quite long when I had my issues.  Here are the results of my analysis.

FloEFD results: Air streamlines around lawn mower blade tips


First, you can see from the above image how the blades cause a lot of swirling air, similar to the airflow around helicopter blades.  We can also look at the velocity contours and velocity vectors inside the lawn mower (shown below).  You can clearly see that at the blade tips the velocity is highest.  This is because the blade is spinning, and the rotational velocity multiplied by the radial distance provides the velocity at that spot, so in general the airspeed will be highest near the blade tips.

FloEFD results: Velocity vectors and contours

Next, I injected the particles at the outer edge of the blade, as only that first bit of the blade will do any cutting (you aren’t walking fast enough for a long piece of grass to make it towards the center of the blade).  From this animation, you can see that the grass particles hit the top of the lawn mower “dome”, than impact a second time along the outer “rim” before going back into the grass.  What you can also see, is that the particles slow down a lot after being impacted by the lawn mower blade.  I think this is the key, in that the slower grass clipping is moving, the more likely it will be to stick/accrete to the top of the dome or to the side rim.  Once some grass starts to stick, the surface becomes rougher, and this promotes more grass to stick, leading to a buildup of grass on the interior.

After sanding some rust off to find the model number of my lawn mower, I was able to dig up a user manual online (a huge surprise).  In it, I found that there are “kick plates” at 4 spots along the top outer part of the “dome”.  From my results, it made perfect sense why those were there.  To try and push down the clippings back into the grass or back into the lawn mower blade, before they could impact the top of the dome and collect.  So I emptied the gas tank and did some searching on my lawn mower, and after removing a BUNCH of hardened, blackened grass, I did discover my kick plates.  I’m not sure if it’s mowing better now because those kick plates are exposed, or that my grass has never been let to grow for more than a week or two.  It still does seem to leave clumps of grass instead of evenly dispersing fine mulch though, so I do think some design improvements could be made, but I’m not sure if this company wants to hear how they could improve their 1992 model.

 One other discovery during my research was that most mulching lawn mower blades have a wing type shape in the interior to create suction to make the grass stand straight up, kind of like that suck-vac haircut machine in the original Wayne’s World movie.  A problem with this though, is this suction can also cause pebbles and other debris to fly up and hit the top of the dome, leading to erosion and holes.  This would also be a perfect analysis opportunity for FloEFD.  As I got my CAD for free on a website, it didn’t have this airfoil shape so it will have to wait for a future installment of my blog.  Still, I think this was an enlightening exercise into another household application where CFD can be used to gain better insight and improve a design.

, ,

26 May, 2011
Welcome to Part IV of a multi-part series, where in this installment Travis continues the “virtual” look at his crawlspace.tm_bio

In the previous installment of “this old CFD house”, we started to peel back the veil on my crawlspace and its moisture issue.  I analyzed the problem as it is, and we made some enlightening findings.  First, really the only way air is getting into the crawlspace is through vents at the front of the house.  If the predominant wind direction isn’t towards the front of my house, my crawlspace will not get the fresh air it needs to get rid of any moisture.  We also saw that the air is stratified in the crawlspace, due to the cool damp nature of this cave-like environment, making it hard for the fresh air to pick up moisture as it travels through the crawlspace.  Lastly, we found the humidity (aka percent saturation) values calculated closely matched those from the crawlspace inspector.


Humidity results with standing crawlspace water

Now that we have an understanding on the problem, we need to start looking into how to fix it.  But not only how to fix it, how to fix it as cheaply as possible, as the bay area real-estate prices are crazy high and the down payment has drastically dwindled my bank account.  While being cheap may get me some ribbing at the office, it is a typical business goal, and FloVENT is an excellent tool to help achieve cost savings for many different types of problems.
After seeing that it takes over 29 hours for a change of air in my crawlspace for a conservative wind speed, I knew getting rid of the surface water had to be the first priority.  If there are puddles of water in the crawlspace a month after the last rain, the ventilation system isn’t working fast enough to remove this water, and disease causing mold can grow in that length of time.  We could have put in a French drain system, but those are costly and would involve destroying our patio, and yet it really doesn’t address the lake we already have in the crawlspace, it just helps prevent new water from getting in (and there are a series of online discussions on their effectiveness and they can be prone to clogging).
In the bay area, there are 3 companies that specialize in crawlspace issues.  2 are actually franchises of the same company, so their prices are the same.  The 3rd is an independent contractor who quoted well below the other company.  Instead of about $14,000 for the 3 phase solution, he quoted 8k.  Still too high for me!  I was interested in getting a sump pump, phase 1 of their 3 phase solution, to get rid of the standing water.   His price was $1000 for this compared to 8k the first company priced this service at.  $1000 is still a good chunk of change, so I wanted to see if that would help fix my problem before committing to it.

A sump pump gets rid of the surface moisture, but it can’t dry out the soil.  So I was still going to have an issue with water evaporating from the ground and needing to get rid of that moisture.  But I hoped that this evaporating moisture would be a lot less then the moisture coming off of standing water.  So I did a little research to determine what a reasonable value was for soil evaporation was.
Well, after reading a lot of agriculture papers on soil evaporation, the best I could find was they quantify this with the term “evapotranspiration” which combines the water loss from evaporation with the water loss from the plants transpiring.  I don’t have any corn fields in my crawlspace, so these rates were likely high.  Also, I don’t have any sunlight down there, which would also drive the evaporation portion of this agricultural moisture loss.  But this was all I could find, so I used the lowest evaporation rate, which was that the top 5-10 mm of the soil will dry up in a day.  Awesome, so with a little more research I found some numbers on the amount of water per cubic meter for different soil types, and I know the surface area of the crawlspace so I could now model this evaporation.
Making the changes in my FloVENT model, I made a fantastic discovery.  By eliminating my surface moisture, the humidity in my crawlspace dropped substantially.  A $1000 sump pump would solve my problem, and I wouldn’t need an expensive plastic liner, or worry about expensive mold problems.   You can clearly see in the following FloVENT results at different heights, that by removing the surface water, the humidity values are reduced.  I was told a safe zone is 50-60% humidity, but even up to 70% I heard is ok.

Results near crawlspace ceiling (house floor)

Results near crawlspace ceiling (house floor)

Results at crawlspace vents

Results at crawlspace vents

You can see there is still a zone of higher humidity, by the garage and behind the porch, due to the poor airflow in this area.  Still this is a whole lot better than before when the entire crawl space had moisture issues.  Also, this is the worst case scenario, with my entire crawlspace soil being saturated with moisture.  The inspectors had only seen standing water in the rear half of the crawlspace, as my backyard slopes toward our house.  As this soil is the furthest away from its moisture source, I have to think it would have reduced moisture content then the 100% I used for this conservative simulation.  This would reduce the humidity in this area.
This knowledge from FloVENT helped me know exactly what was needed to solve my problem, and saved me a lot of money by avoiding the costly repairs that I was being told I needed.  Only by understanding your problem, whether it’s a lake under your house or any number of design issues engineers face every day, can you really begin to work on a solution.   FloVENT simulations can help not only with the understanding of the problem, but also by investigating different possible solutions, and in do so help you find the most cost effective solution in the most cost effective manner,  by utilizing your desktop computer instead of costly physical tests and prototypes.

Join me next time when I investigate another house hold problem with one of Mentor Graphic’s CFD products.

, , , , ,

19 May, 2011
Welcome to Part III of a multi-part series, where in this installment Travis takes a “virtual” look at his crawlspace.

For this next installment of “this old CFD house”, and really the entire reason I started this blog series, I’d like to take a look in my crawlspace. Now, from whatm_biot I understand, Feng Shui says running water is good for wealth. Well the opposite is true when you have a lake beneath your floor. In California, houses are built on posts and are a certain distance above the ground, and the space in-between is the crawlspace. Our crawlspace (it’s officially ours now), has a bit of a moisture problem, with standing water and high humidity. It’s this high humidity that can cause a world of problems, from mold to rotting your floors.

In fact, during the inspection process we had a crawl space specialist look at the problem. After his inspection, he gave his assessment on the issue and what it would cost to remedy. He gave 2 options, or really, a phase 1 and phase 2 (a “how much do you love your family type test). I like to relate expensive costs to Ford Mustangs, my favorite car, as it’s a good way for me to evaluate what I would rather spend that amount of money on (it’s usually a Mustang). So, to get a sump pump installed with some trenching, and a plastic barrier to eliminate soil evaporation, we were looking at a low end Mustang, V6, cloth interior or maybe an older model. I was pretty sure I could dig a hole, install a sump pump and put some hefty trash bags on the ground for a whole lot less than that. But of course, I was worried about being penny smart and pound foolish, so I knew I needed to analyze this with FloVENT to ensure whatever course of action I took it would fix our prospective mushroom garden under our house.

The main issue was that our backyard was sloped towards our house, and re-grading was an even more expensive solution. Another possibility was running a French drain around our house, and destroying a lot of concrete patio along the way. The problem with that is it doesn’t help in get rid of the current moisture in the crawlspace. This brought me to my first FloVENT analysis.

We frequently heard that our house had a greater than average number of vents into the crawlspace, more than double the current code, which I believe, is one vent for every 150 sqft. This should provide more than adequate ventilation to remove any humidity, but yet we have 88% humidity (well above the safe 50-60/70 zone) and standing water a month after the last drop of rain. Seems like all this supposedly good airflow under our house is doing is making the floors super cold and driving up my heating bill. So, like you, I wanted to see how effective this “venting” strategy is at getting rid of moisture. I’m sure in the 1960 or 70’s or whenever venting became the standard way of building a crawlspace, no experiments or CFD was used to see if this design accomplished it’s design goal of a dry crawlspace, so I needed to take the initiative and analyze mine to get a handle on this problem.

I created the FloVENT model geometry based on the blueprint layout of my house. I eyeballed my vent positions and looking at the vent style (shown below), I estimated it was about mesh80% open to airflow for my air resistance. Initially, I didn’t model any obstructing geometry in the underfloor, as I didn’t know the location of ductwork. After running the analysis though, I did start to wonder how much obstructions would affect the performance of my crawlspace system, particularly floor joists which I thought would “trap” any natural convecting air up at my floor. I didn’t go into the crawlspace (…spiders…) but I know the size of my ducts and where they start/end, and I know the size of my joists and their spacing. Looking at the results, it only changed my air change rate by less than an hour, so not a huge effect, but I’m getting ahead of myself.

aspect of this problem is the boundary conditions. In my case, my house is basically as wide as my lot, with fences on the sides. So, there isn’t going to be any air getting into my crawlspace from the sides with any type of wind momentum to penetrate any significant distance into the crawlspace. Similarly in the back yard, in California land is at a premium, so back yards are small. Any wind blowing from that direction will have to go over my neighbor’s house and our fence and in my opinion unlikely to dive down and into my crawlspace. Right there, I was starting to see the issue; only the front of my house really had the ability to have any airflow get into my crawlspace, simple as that. Unlike a roof, where you can harness natural convection to drive air out the top of the roof, a crawlspace has no height change between vents and is basically a dark cave with no heat source to create that convection current (which would get caught up in the floor joists anyway). No, it needs a wind, and if your house is like mine, a good half of the front is taken up with a garage. That garage, takes up ideal ventilation space. That means that the air entering in the my only front 2-3 vents now has to drive the air in the entire crawlspace, and make a pretty good 90 degree turn to clear out the dead air space behind the garage/front porch. Oh, and this is assuming that the wind is coming towards the front of my house for a substantial amount of time (which I had to assume for the analysis).

houseNow that wind is a pretty important component of this equation, not just from an airspeed point of view, but also from a temperature and humidity aspect as well. Yes, we need to try and drive out humid air with air that also has some humidity to it as well, a less than ideal situation. Luckily, the weather underground can provide such stats, and it seems like living next to the San Francisco bay, the air is usually around the 50% humidity range. For wind speed, unfortunately there isn’t a weather vane in my front yard, and as I’ve discussed direction is just as important as speed, so I went with a conservative 1 m/s which is the bare minimum air speed a person can feel on their skin. While you may notice the breeze and gusts more, I’m hoping and praying that there is at least a slight, unnoticed, constant wind blowing across my street and into my crawlspace.

Lastly, we need to model that evil humidity. I’ve modeled pools and things before, but that math assumes things like a temperature difference between the air and pool/hot tub water driving some evaporation, and a correction for kids splashing, none of which I could use for my sub-residential ocean. Believing that this water would behave like a humidity “no slip” condition, I modeled it as a 100% humidity source at the surface of the standing water. From 3rd party accounts (again, spiders and deadly mold down there), I’ve heard the water was over about half my crawlspace, but that was one moment in time. Maybe in the worst case, it could have filled my crawlspace, so I modeled it as such.

Below are some of the results. First up are 2 plots of the humidity at different elevation above the ground. Comparing the 2, you can see the substantial difference in humidity in just 1 foot change in height, indicating pretty stratified air. You can see that for the most part, the lower crawlspace area is a pretty constant humidity of 80% and above, which confirms what the 85% humidity meter reading from the inspector (which I didn’t fully trust as he has a financial interest in showing I have a problem). Still, it’s pretty depressing to see all that humidity. The dark blue areas in the images are the garage, the concrete front porch, the chimney, and the heating ducts. You can also see the plumes of humid air coming out of the vents around the house.


image_02Looking at the speed contours, you can see that really only one of the vents is providing a nice jet of air, but for the most part the airflow in the crawlspace is non-existent (for comparison, a natural convection current is ~0.2 m/s). Oh ya, and did I say those vents are right behind a empty, but soon to be filled, flower bed and some future hedges (see previous image)? It couldn’t be in a worse location…


Sometimes it’s easier to understand the airflow by looking at streamlines, so I tracked the airflow from all the inlet vents to see where the air is going in my crawlspace. It looks like I have a nasty looking vortex in my underfloor, a big issue with raised floor data centers, but for a crawlspace more an indication of poor airflow.


Clearly, the right side of my house gets the best airflow, and behind the porch and garage the airflow is a lot worse. Also, you can see from the humidity (percent saturation in this chart) that the moving air doesn’t pick up much humidity before leaving the system. When looking at the results in 3D you can see the airflow hugs my floor (it’s ceiling), likely due to the fact the air in the crawlspace is cool and denser then the fresh wind air, resulting in a stratified air in this space. Comparing with my original FloVENT analysis without the HVAC vents, I found that the HVAC vents do help create some mixing effects and reduce the average humidity ratio in the crawlspace, an interesting finding.


For me, to really understand the magnitude of the problem, I needed to look at some numbers. I figured the real design goal of these vents is to achieve a certain amount of air turnover in the crawlspace, so I wanted to calculate this air change rate to see how long it took to completely replace the air in the crawlspace. The calculation was easy. FloVENT tells me the volume air flow rate in all my vents, and I knew the volume of air in the crawlspace (the square footage of my house times the height of the crawlspace), so with some simple math I calculated a air exchange rate of about 29.5 hours. Now I could see why standing water could exist for so long under my house. I’m not sure if this is out of the norm for crawlspaces, but clearly, air can only hold a certain amount of water, and if it takes more than a day for the air to leave the house, I could easily see it taking many weeks to dry out the crawlspace after a good hard rain, especially if the air is stratified and the new air really isn’t interacting with the moist air. Of course, you can look on the bright side and say that the denser, moister air is down near the floor of the crawlspace, separated from the floor by the incoming air. Therefore, it’s kept from rotting the wood of the floor or growing mold there. But, I don’t want mold growing anywhere, so the moisture issue has to be solved.

Now that we understand the problem, we can start to investigate my solution, and whether it will accomplish what we want, a clean, dry, mold free crawlspace. So join me next week for the conclusion of this blog installment.

, , , , ,

29 March, 2011
Welcome to Part II of a multi-part series where, in this installment, Travis takes a look at a typical chimney.

tm_bioFor this next installment of “this old CFD house”, I’d like to investigate another gremlin that was discovered during the inspection period of the house we are purchasing. This one related to the chimney. The current damper is rusted and hard to operate. According to the chimney expert we had out to estimate the repair, the damper doesn’t fully close due to all the corrosion. He stated that at a minimum, 8% of our heat would be wasted, going up the chimney which he said for the average home works out to be about $200 a year. I instantly had a flashback of myself as a kid leaving the door open and having my dad yell about “not paying to heat the neighborhood”. I as well don’t want to “pay to heat the neighborhood”, but I also don’t want to pay the chimney guy. So I needed to decide what was worse, leaving it as is and wasting some heating energy, or paying the $600 to have this fixed.

His statements gave me a lot of questions. Where were these “average” houses? I’m sure the heating losses would be worse in Minnesota then California, where we had only a couple months that need heat and the outside temperature is rarely below 32 degF. He said the damper doesn’t close completely, but how close to being closed is it? I would think a ¼ inch gap would leak a lot less than 8% of my total heat, but I don’t know. I also don’t know how much worse the heat leakage is as the opening increases.

I knew I needed to use FloVENT, our HVAC CFD tool to solve this problem, but I had to do some research. What does a typical damper system look like in a chimney, and what does the inner workings of a chimney look like? Well I found this picture on a chimney company website. It seems to exhibit some of those dreaded “airflow arrows”, which I hate because usually the airflow doesn’t do anything like that. For a future blog, I may look into the air flow of the fix, the “Strong Draft” side of the picture below to see if FloVENT matches the image and if the airflow is any better between the old and new. I already have questions about whether a strong draft is better, as if it’s winter and I’m heating my house, do I really want MORE air leaving my chimney? I likely want the minimum amount of airflow needed to remove the smoke from the house. But let’s get back on the current topic.

With an idea on the geometry, I needed to research one other piece of information. I needed to find out what pressure was realistic for my house. I found this image and article from my former employer, the National Research Council of Canada, which showed that about the worst I could expect for my home was for it to be pressurized to about 10 Pa. I didn’t use the open chimney, as I don’t know how “open” the criteria is for an open chimney system.

With that, I created the FloVENT CFD simulation model. I ran it with a 68 degF inside temperature (72 degF? that’s what sweaters and blankets are for), pressurized to 10 Pa, with 32 degF outside air temperature. The living room was assumed perfectly insulated so I could focus on the heat loss due to the faulty chimney damper. I used our Command Center application to automatically run through a sweep of damper gaps at 0.25 inch increments to investigate how the heat loss changes as the gap changes.


So from the simulation I found how much energy is leaving my house, but I need to associate a cost to this. As I haven’t gotten a utility bill yet, I don’t know what the typical energy usage is at this place, or what the cost of natural gas is. I researched and found that $4/CCF (100 cubic feet) seems to be reasonable for the price of natural gas, but I need to figure that cost out in energy units. According to one site, 1 cubic foot of natural gas is equal to 1000 BTU’s, so 100 cubic feet is 100,000 BTU’s and costs roughly $4. Using this, I obtained the result table below.

chimney_tbl_01After starting to do the cost calculation, I realized that using 32 degF for 24 hours a day for 6 months was way too conservative for California.  If I ever move anywhere with a cold winter, these numbers in red will be seared into my brain if I have any type of chimney.  I adjusted my cost calculation to just the overnight hours for 2 months, as that is seemed more reasonable for California.  Still, it’s neglecting the day time heat loss, which I still need to address.  I can see that a mere 2 inch increase in the gap results in over a hundred dollars more in heat loss and is approaching the quoted $200 from the chimney repair man.       

To address the issue about the heat loss during the day, I ran another Command Center sweep on the different gap sizes, but with a 50 degF outdoor temperature.  This temperature just seems right to me, as on cold days here the temperature highs are around 50-60 degF.  I looked at the cost 2 ways, one, using 24 hours a day and 6 months, mainly to compare to the 32 degF and see the magnitude of the buoyancy forces on my heat loss.  Secondly, I calculated the cost for 16 hours for 2 months and added that to the overnight 32 degF heat loss cost to obtain the total 2 month winter costs.



The 2 major things to take away from the above findings is that with more realistic temperature estimates, the chimney inspector estimation for heat loss cost was very reasonable.  I’m not sure how long it took him to run his CFD analysis though Cheerful.   The other main point that’s illustrated in the graph and the red table row is how much of an effect an 18 degF temperature change makes on the heat loss.  Partly due to the higher buoyancy forces driving air flow up the chimney and partly due to the simple fact that larger temperature differences drives more heat transfer.  It’s clear to see though, that the slopes of the graph lines are different, with the low temperature line having a steeper slope.   So in colder climates, the heat loss of a house is more sensitive to damper gap size.

Looking at those results, and thinking about buoyancy forces, made me think about my pressurization effects.  Right now, the simulation had 10 Pa for my house pressure.  Is this driving all heated air out of my house, or is it a non-factor?  It’s hard to have a feel for Pascal’s, but just thinking that atmospheric pressure is 101.3 kPa, makes me think 10 Pa isn’t a big factor.  I could likely have calculated the cross section area of the chimney opening to get an idea of the real forces involved.  But, I took the easy route, changed the 10 Pa to 0 Pa and hit “GO” on my 32 degF FloVENT simulations again and let it run during my lunch break to see if having an unpressurized house would have substantially less heat loss through the chimney.



I didn’t expect this.  Reducing the pressure, which seemed pretty low, down to 0 Pa reduced our heat loss by over half.  This effect is as important as the outside temperature and the damper gap.  While an unpressurized house likely isn’t feasible or desirable, reducing the pressure of the house could yield some cost savings.

So after looking at the results, I think it’s clear that fixing the damper is a sound investment.  For a gap of anymore than 1 inch, it will only take about 3 winters to get a return on investment on the $600 repair if our home is pressurized to 10 Pa.  Of course my calculations assumed that every day/night beyond 2 months was the 72 degF weather that makes California so expensive to live.  Any additional inclement weather will reduce that payback period.

Join me next time when I look at moisture in my crawlspace and see if my under floor ventilation is up to the task.

Travis Mikjaniec

, , , , ,

19 February, 2011

When I saw that Robin had written a blog about a beer fridge I thought he had beaten me to the punch. In my spare time I too have been using our CFD tools to analyze a beer fridge. The differences are I am using FloVENT  and my fridge is a self-service refrigerated showcase.  The story of the showcase is not complete but I want to provide the first installment.

The idea of the open (self-service) refrigerated showcase interests me as a thermal/airflow designer and as a consumer.  I certainly like standing in front of a showcase without having to contend with an obtrusive door.  Invariably though, while I am weighing the beverage options, I think about the design of the refrigerator.

I would like to believe that a “jet” of air from the top of the unit will provide a curtain between me and the beverage.  It has been my experience that though you can sketch the flow path on a napkin that you desire air doesn’t always know where you had intended it to go.  How much of that curtain is reaching it’s destination to be re-chilled?  How much of it enters the aisle?  Does it cause discomfort to the consumer?

To begin to understand the airflow path I built a representative FloVENT CFD model. My display has two supplies at the top, the outer supplies air at about 1.5 ft/sec and the inner at half of the outer. I wanted to pick some very low airflow speed that is essentially undetectable by a human, because I never feel any air movement in these “curtains”. The two streams of air, where only the inner stream is re-chilled, are imagined to travel to the returns near the bottom of the display. To show the air flow path I have uploaded an animation shown below, where the particles are colored by temperature.

To my delight it appears that even a curtain at a very low flow speed provides a reasonable barrier to the customer. For further understanding I seeded the inner and outer supplies so we could quantitatively determine the effectiveness of the curtain. The contour plots below are colored by the flow markers for the inner and outer flows. A value of 1 indicates 100% flow from the associated supply.

The initial analysis shows that the chilled air is contained pretty well but some of the un-chilled  air does enter our display.  The next step will be to vary some of the flow and design parameters and perform a Response Surface Optimization to determine what variables influence the performance the most.

, , , , , , ,

10 February, 2011
I’d like to introduce my first guest blogger, Travis Mikjaniec. Travis is a colleague of mine who is entering the world of home ownership. Travis has questioned some of the recommendations from the various home inspectors. In a multi-part series Travis will debunk or otherwise quantify the value of the recommendations.

Starting the process of buying a house is a daunting task.  tm_bio Beyond all the forms and stress, are all the inspections.  I’m learning there are many household issues that relate to heat and airflow.  As a CFD engineer, and a soon to be cash strapped home owner facing a laundry list of repairs, I wanted to prioritize and investigate the various issues using the various CFD tools at my disposal.  So please follow along with me while I use some company time analyzing the various ills of my property in my series of blogs that I affectionately call “this old CFD house”. 

 For my first investigation, I’ll be looking into a problem discovered during the general home inspection.  In California, most houses don’t have basements, I think due to earthquake concerns.  The houses are 2-3 feet above the ground on piers/posts and have a perimeter foundation.  This under floor space is known as the crawlspace, and seems to be a common area for household issues, likely because it’s out of sight. 

 Our issue is related to the heating ducts.  In northern California it does get a little chilly in the winter months, so heating is still a concern.  Our furnace pushes air downwards into ducts in the crawlspace that then distributes the air throughout the house via floor registers.  According to our home inspector, the ducts should be 4 inches above the ground.  Ours??  Directly on the ground in many spots.  If you’re like me, you’re instantly thinking about how much worse heat loss from conduction is to convection.  I learned in an army first aid class that one blanket under someone is worth 5 on top, but I’d never use this rule of thumb to infer that raising the heating ducts will decrease my heat losses by 5 times.  Convection is highly dependent on airspeed so impossible to use general rules of thumb to estimate it in a complex system.  Therefore I was interested in quantifying this heat loss using CFD.  As I didn’t know the exact layout of the heating ducts, or the conductivity of the crawlspace soil, or the exact CFM or BTU’s of my furnace yet, I had to do some estimating.  I created the entire house as a “system level” model, show here. 



 From that, I created a zoomed in model with a simple duct from the furnace room to the family room as an estimated duct line, shown below.  I researched and found a typical CFM for a furnace is about 1000 cubic feet per minute, and estimated there would be about 1/5 of the air going towards the family room/breakfast nook/kitchen area, so the airflow through this duct was 200 CFM.  I defined a set inlet air temperature of 74 degF at the start of the duct, as it would make for easy comparison by simply looking at the final temperature of the different cases.  Then I ran 4 cases in FloVENT, our HVAC CFD software:  a bare duct conduction, 2 inch fiberglass insulation conduction, bare duct 4 inch airspace, and finally 2 inch insulation 4 inch air space.




As shown in the table above, the world of thermodynamics is a strange place.  Whether the duct is on the ground or above it, as long as it’s insulated, it doesn’t matter much.  Because the conductivity of dry soil is pretty low at 0.36 W/m2K, it doesn’t conduct much.  Its conductivity isn’t that much different than the conductivity of air at 0.026 W/m2K.  Of course, this depends on a lot of the estimates I made initially.  If the soil conductivity increases substantially, say due to moisture, then conductive heat transfer will increase.  And, this is just a heat transfer point of view as we are ignoring key factors like if the insulation gets damp or degrades.  But we also estimated the ductwork was in constant contact with the ground, a worst case scenario.  Still, I know the ducts are wrapped with insulation, and the winter is almost over, so I’d like to procrastinate on this chore till another day, and now I have the results to prove it when my lovely lady asks me to get in that dark, muddy, spider filled crawlspace to fix it. 

Join me next time when I investigate my leaky chimney flue.  Home energy efficiency nightmare?  Or a non-issue, hyped by the chimney industry’s marketing machine?   

Travis Mikjaniec

, , , ,

29 December, 2010

I’ve been a iphone user for about a year and a half now and for the most part I am a satisfied customer. I do need to schedule a visit to the Apple store because lately my 3 month old iphone 4 operates about 4 hours between charges.  Luckily I have two chargers in my living room, one in my truck, and one in my cube at work but this situation doesn’t seem sustainable.  The beauty of owning an iphone, or any Apple product I suspect, is that there are a number of local stores and that they provide exceptional customer support.

My wife just got a iphone 3G and she too is pretty satisfied but it does make her long for the Sony CLIÉ she used to use.  She loved the To-Do tracking ability.  I thought I would do something sweet and gift her a To-Do app for the iphone. (I wanted to start that sentence with “So” but a couple of years ago a British friend of mine commented that all Americans start sentences with “So”…this is me bucking the trend).  As part of my Christmas from my wife and kids they bought me a itunes card which I promptly redeemed before I had a chance to lose it.  With plenty of funds and my heart in the right place I gifted her an app.  Well, and here is where my rant starts, it would seem that you need to use a credit card (or Paypal) to gift an app.  I have $$ in my itunes account but for some reason it wouldn’t allow me to use them.  It could be that this is possible and I am just doing something wrong.  I did plenty of clicking in itunes and google before I succumbed and put the $1.99 on my credit card (what a guy).  It could be that Apple hasn’t developed the technology to add that button in itunes.  It could be a way for them to see if they could squeeze blood from a turnip.  It could be a test to see if your heart is really in the right place, after all me gifting an app from funds I got from a gift card could be considered “re-gifting”. 

Whether it is possible to gift without a credit card I can’t say definitively, I wasn’t able to do it today.  It could just be a well guarded secret.

, ,

23 June, 2010

CFD is a powerful design tool when considering a data center cooling scheme. With CFD you can display contour plots of temperature, speed, and for the purists there is pressure.  I, perhaps naively, have always considered pressure to be a means to an end.  I can absolutely see the value of pressure in terms of “pressure drop” but for me to look at a contour plot of pressure at a room level and deduce how my design should change makes my head hurt.  Don’t hate me.   You can also animate the air flow by seeding the flow with neutrally bouyant virtual particles.  Very powerful indeed but qualitative in nature.  In other words, very useful in painting a picture of the quality of the design to colleagues or management but doesn’t give the designer the full analytical benefit.

To illustrate the value of the various types of CFD outputs for a data center airflow design consider the following scenario.  We have a raised floor data center with two downflow room coolers.  Each cabinet has a floor tile adjacent to it with the hopes of supplying a dedicated source of cool air.  The image below, a temperature contour and velocity vector plot near the inlet to a row of cabinets,  illustrates the result of this design.  Notice that the air in the upper portion of the cabinets is heated prior to entering the cabinets.  Upon closer inspection we also notice that the first two floor tiles are supplying air to the plenum rather than delivering cool air to the caibinets. Not really what we had hoped for.


Another useful output from CFD is the ability to animate the flow which I have shown below.  It shows, quite clearly I believe, that we have some flow sneaking over the top of the cabinets rather than heading straight back to the cooler.

A relatively new metric when designing an airflow management solution for a data center has to do with this thing called “Capture Index” (James W. VanGilder, Saurabh K. Shrivastava: Capture Index: An Airflow-Based Rack Cooling Performance Metric, ASHRAE Transactions 2007, DA-07-014, Volume 113, Part 1)
It is a very powerful metric which is worthy of a dedicated blog, and a whole lot more.  In my next blog I will illustrate the additional information which can be gleaned from this design by way of the “Capture Index”.

, ,

17 June, 2010

Just a quick note that I will be participating in an EDN Editorial Webcast on Friday, June 25, 2010.  There are four panelists that offer up some valuable insight into the challenges associated with the reliability of LEDS.  LEDS are a relatively new technology when compared to other semiconductor devices and standards for testing and reliability prediction are not yet universally accepted.

You can sign up for the webcast here

, , , , ,

28 May, 2010

I really enjoyed both of the Web Seminars on heat sinks; Heatsink 101 presented by Alexandra Francois-Saint-Cyr and Heatsink 201 presented by Dr. John Parry.  I was especially interested in a couple of equations presented in Heatsink 101 that estimated h (convective heat transfer coefficient) and minimum fin spacing.  The equations were derived from a Flat Plate in Parallel Flow theory but were simplified so that all of the fluid properties and unit conversions were already taken into account.  This inspired me to carry it a bit further, with a few more assumptions.  The idea behind these formulas aren’t really to allow the design of a heatsink but more as a preliminary sizing tool.  With this in mind I wanted to come up with a single formula that would estimate the required volume to generate the desired heatsink thermal resistance for a given velocity.  Enough of the work, I think, is shown below.


Aside from some simplifying assumptions on the heat sink surface area was the idea that the length, width, and height of the fin area were all equal.  None of these dimensions are related to the actual fin thickness, this calculation doesn’t go into that level of granularity.  Essentially it assumes that the heat sink is at a single uniform temperature where the flow between the fins never fully develops.  The number of fins is based on the assumption that the flow never fully develops;  The fins are spaced at 2δ apart at the distance L along the plate.

A few years back I wrote an excel vba macro to estimate heat sink performance that took into account such things as , airflow bypass, heat sink thermal conductivity, heat spreading, and fin efficiency.  The problem was that when I was in a situation where it might be useful I didn’t want to populate it with all of the information it required.  I have started to dust it off again though with an attempt to streamline the input.  Ironically, I guess,  it uses a Nusselt number based on internal flow in a rectangular channel with an infinite aspect ratio where the flow is fully developed.  The approach presented here is a bit more useful, at least to me,  because it requires only two inputs.final_equationAside from checking for consistent units and producing the results in the table below, which I think look reasonable, I haven’t done any testing.  For better or worse I do think this is an equation I will use for a first pass size calculator.  This table shows what the estimated L,W,H needs to be “roughly” to achieve the desired heat sink thermal resistance (10→1) at various velocities (0.5→5)


I encourage you to let me know what you think of this or if you think I made a mistake.  Also,  if there are any other approaches that balance the required input (minimal) to accuracy/usefulness (somewhat) I would very much like to hear about it.

, , , , , , ,