Travis Mikjaniec's BlogTravis Mikjaniec's Blog RSS
This is a bit of a follow up to my previous blog about the FloVENT analysis of the Mentor Graphics Data Centers in Wilsonville, OR and Shannon, Ireland.
Well the one in Shannon, Ireland has been online for a while now, and our Wilsonville, OR data center is scheduled to come online December 2012. Of course, it will take some time to start shifting equipment and IT load to the new data center from the smaller sites across the US. But, within a short amount of time we will hopefully have some real data to compare to the simulation results FloVENT had provided us. To me, working for a software company, in the mechanical analysis department, getting real data to compare simulation results to is the hardest thing. Any experimental data I see is from customers, and is always proprietary. This is one time where the experimental data is ours, so I think it is exciting to be able to bring our design and analysis full circle by seeing some real world results. It should make for some good blogs/papers/articles in the future, so stay tuned.
Related to this data center project, recently I was asked to present at the ASME chapter in Dallas Texas, and the vice chair thought Data Centers design would be a good topic. Luckily I had my presentation from Semitherm on our data center design, so I have plenty of material to talk about on this. If you live in the Dallas area and are a ASME member, come check out my talk.
The talk is next week, on Tuesday January 8th, in Plano Texas (3115 W. Parker Rd, Ste 400). The meeting starts at 5:30 pm and my presentation is scheduled to start at 7 pm.
I hope to see you there!
Previously, I co-authored a semitherm paper on the use of our HVAC CFD tool, FloVENT, on the design of a Mentor Graphics data center. Recently I have seen this topic republished in the data center journal with the title “EDA Vendor takes its Own Advice on Data Center Design”. I find this topic interesting, so I thought I’d blog about this and break from my typical CFD blog topics on household applications.
To give you the background, over the years, to support Mentor’s growing needs, individual offices had created their own small data center rooms. Mentor saw a way to improve capacity, efficiency and redundancy by consolidating these small data centers into a large data center. It just so happened that around this time, Mentor Graphics had just acquired Flomerics, so we now had some internal capabilities to do CFD analysis on this project using FloVENT.
I’ll skip a lot of the details that have already been covered in the semitherm paper and published elsewhere, and just focus on the design progression. To start with, the initial design concept was for a raised floor design. These are very common in exisitng data centers. By having the air conditioned air put into the raised floor plenum, you theoretically allow it achieve a uniform pressure distribution and it allows you to control where the air is supplied to the equipment using perforated floor tiles. Noticed I said theoretically. In our case, we stared with a 2 foot raised floor design, but besides the extra cost of $35 a square foot (~$100,000 total), the FloVENT simulation showed there would be pressure problems and the raised floor plenum had to be bigger. Increasing the raised floor height to 3 feet would increase the cost from $35/sqft to $40-45/sqft. In comparison, the initial cost of ductwork for a ventilation system was only $10/sqft, so the raised floor design was abandoned.
The next design contender was a dropped ceiling design, where a chimney system would duct the hot exhaust air from the cabinets to the dropped ceiling plenum. Here, the air would be allowed to achieve a uniform pressure before being ducted to the roof top air conditioning units. There were 2 main issues with this design. A simple FloVENT simulation indicated the plenum would need to be 10 ft deep, but as the building was only going to be 20 ft high, this would drastically cut the data center air volume which would have a negative impact on the supply air uniformity. You can see from this from the 4 FloVENT result images below. If the supply air is not uniform, then you are going to have racks that are either getting hot air, which is bad, or not enough air, also bad. The 2nd issue related to fire safety, as a return plenum of this size would need to be metal lined to meet fire codes, and the data center could be shut down if the fire marshall was ever not happy with it. So this design was also abandoned.
We also looked at hot aisle containment, where the area that the cabinets exhaust to is sealed with doors. The issue with this design is maintanence of the equipment. OSHA has regulations about how long someone can work in a area at a certain temperature. As some servers have a 54 deg F temperature rise, the contained hot aisle would have temperatures in above 110 degF. The OSHA regulations for this harsh environement made this option not cost efficient.
The other design option was a chimney system, so that a technician would only be exposed to some exhaust air but would be working in the cooler supply air in the room. Because this averaged air temperature was substantially cooler, the OSHA regulations were not detrimental to the design. Below is an image of multiple cabinets being serviced at once. FloVENT showed that even when the equipment was being serviced, 2/3 of the exhaust air continued to travel through the chimney system. This was the winning design concept, to utilize exhaust chimneys to duct the hot exhaust air away from the servers.
Even with that though, there was still plenty of FloVENT analysis to do to iron out the design details. One example of this was the design of the collector duct. Basically, the chimneys from the equipment feed into a collector duct, which then feeds a return plenum, that supplies air to the roof top air conditioning units. What FloVENT found was if the return plenum was at one side (left in the image below), it causes a substaintial pressure gradient along the collector duct, so that the cabinets at the opposite end had air flow issues. This was corrected by moving this connection to the middle of the collector duct, which substantially reduced the pressure gradient some servers had to work against.
I hope this helped show a real life design that utilized CFD analysis as a intrigral part of the design process. Without using FloVENT to analyze this data center, we would have not came up with a efficient design. For those of you interested in more details on this project, the title for the semitherm paper was “Data Center Design Using Improved CFD Modeling and Return on Investment Analysis”. You can download a copy of the paper from the IEEE website.
When I watched the opening ceremonies of the Olympics, I knew I needed to analyze something related to the Olympics during this special time of year. At first I thought I would analyze the Olympic Rings. Not a real scientific CFD analysis, but just something cool and easy to do, by hollowing out some Olympic CAD rings and putting some flow through. Well, using FloEFD to do this analysis was easy, getting this by our legal team was not so easy. Term’s like “copyright infringement” were thrown around, and since I didn’t want to worry about job security I decided to look for something else to analyze.
I went to my favorite free CAD website, grabcad.com. I found some different sporting equipment to analyze, but the one that stood out to me was a badminton shuttlecock (hearby referred to as a birdie to keep this blog semi-professional). I’m not a badminton fan, but my wife and my cubical neighbor are, so I thought they would get a kick out of this. Also, compared to all other sports, badminton isn’t played with a ball. The birdie is designed to be very stable in flight, but also to be very “draggy”. This meant there would be some interesting aerodynamics at play.
I started investigating the CAD, and noticed the dimensions were very wrong. It was a skyscraper of a birdie. So I adjusted the CAD to be the right dimensions based on what I could find online, and started to analyze the results. When I looked closely at the drag numbers, I noticed they were extremely large. It seems, that I had missed one important detail, the units. The values for the dimensions were correct, but my PTC Creo model was set to inches instead of millimeters. I guess just like the wood shop motto “measure twice, cut once”, I needed to have a CAD-FloEFD motto “measure twice, extrude/revolve/pattern once”. Or something like that. Luckily, because FloEFD for Creo is in the CAD tool, all I needed to do was update the units. The CAD file updated, and FloEFD recognized the CAD change and I could simply run the existing analysis setup on the updated geometry. A huge time saver.
Now, at this point the Olympics are rocked by the big badminton controversy. What are the odds the sport I choose to do a FloEFD analysis on became the black mark on the games. Anyway, I decided to analyze the birdie at 2 speeds. Online, I found the maximum speed a person has smashed a birdie was 206 mph. Think about that. A slapshot goes about 100-110. A fastball pitch in baseball is about 90. Even tennis serves aren’t near that speed. Now, the thing the website didn’t say is was this the average flight speed, or was this a instantaneous speed of the birdie right when it was hit, which then slowed immediately. I am 99% sure a birdie isn’t flying at 200+ mph across the length of the court, but I thought this would show some cool aerodynamics.
My other flight speed used was 60 mph. This was my initial guess for a typical speed for a birdie, before doing the above research. So let’s just call this one the speed of a hit birdie when you are trying for a better semifinal match in your round robin
Let’s start by looking at the 206 mph case. First, I always like to show some results showing FloEFD’s adaptive mesh. That is what makes analyzing the complex geometry of a CAD model possible. You can see how not only it refined on the geometry, but also downstream in the wake of the birdie.
We can see behind the cork/nose, there is a large recirculation zone. We can also see how the feathers push the air outward as it flows away from the birdie. But, if we look at the velocity vectors, we can also see how because the feathers are on sticks/quills, this allows some space where air is sucked into the middle of the birdie. It is sucked down because of the low pressure from the recirculation zone. You can also see on the inside of the feathers there is a low pressure area. I believe this is because of the air being sucked into the middle of the birdie, it has an “angle of attack” compared to the feather, and there is a flow separation. I think this explains the flight stability of the birdie. On all sides there is a inwards force on the feathers, keeping them from yawing. Of course to really analyze this effect, we would need to analyze the birdie in a yawed position, and see the resulting aerodynamic moments. That is an analysis for another day.
Looking at the drag force on the birdie, we see it has 4.4 N of force on it. This may seem small, but remember the units of a newton. 1 N is the force to accelerate a 1 kg object at 1 m/s^2. Now our birdie doesn’t weigh 1 kg, it is more in the 5 g (0.005 kg) range. Therefore, at 4.4 N, our birdie is experiencing a deceleration of 880 m/s^2, or ~ 90 g’s. I believe a fighter aircraft can pull about 10 g’s, just for comparison. This tells me for sure that there is no way the birdie was flying through the air at that speed for any amount of time. Instantly this drag force would be working to slow the birdie down. Of course if the birdie is only in the air for a second or two, then there isn’t a lot of time for this force to act, and the birdie would still be moving at a good clip when it reaches the opposing player.
Looking at the 60 mph case, the flow field looks the same except for lower velocity values. The force is a lot lower too, at 0.144 N. Still, that equates to a deceleration of 28.8 m/s^2 (~ 3 g’s), but well below that of the world record speed.
In the future, it would be cool to analyze the birdie in a yawed situation to see if there is a righting/stabilizing moment. Or even more interesting to look into the forces on the birdie right when it has been hit, and is flying backwards. What are the aerodynamics at play that spin the birdie around to fly in the right orientation. I’ll have to remember to do that in 4 years at the next summer Olympics.
I know I promised that the next blog would be a BBQ CFD analysis, but I came across this model and it was too interesting to pass up. A number of years ago, I was flipping through the channel guide and saw a show about choppers. At first, I thought I read copters, and being an aerospace engineer, watching a show on helicopters was something I was interested in. Of course the show was about motorcycles, but the first episode I saw was of a fighter jet inspired chopper. From that point on, I was hooked on all these types of motorcycle shows.
The other day, I came across this CAD model on www.grabcad.com, of a person’s dream chopper. I wondered about the aerodynamics of a chopper, as just a couple weeks earlier I had seen a “Mythbuster” episode where they tested the fuel efficiency of a motorcycle vs a car. The motorcycle killed on fuel efficiency, but was poor on air pollution. Then they made an aerodynamic shell and tested it again to see if they could improve the fuel efficiency enough to overcome the bad air pollution. Interesting stuff for sure.
Of course the motorcycle they used was nothing as cool as this. I’m sure unlike mass manufactured motorcycles, a custom chopper has never been wind tunnel tested or analyzed using CFD…until now
For my analysis, I set the airspeed to be 75 mph, because just like the song, “I can’t drive …. fifty five”! Other than that, the setup was pretty easy. I just needed to figure out the RPM for the wheels at 75 mph, and I took an educated guess at the engine exhaust/intake air flow rates. I created some CAD to represent a two lane road (I couldn’t figure out how to paint a centerline on my pavement texture map), and some grass. I didn’t get to making a CAD person riding the chopper, as then I’m sure most of the drag would have to do with his riding position (leaning forward vs straight up) and really what I was interested in was the bike itself.
Here are some of my results. Now, the bike looks fast even just sitting there, but in this first cut plot showing air speed and the mesh, doesn’t it look like it’s going even faster? If we look at the second contour plot of air speed, this time cropping out any air speed faster than 70 mph, you can see how far downstream the wake of the chopper extends. You can also see at some point, the wake becomes unstable and oscillates, creating that wiggle at the end. But I also like this plot because it shows that the wake also is very compact in the width direction. If you were a person on the grass, you wouldn’t feel the air move much when the motorcycle passes by, unlike when a transport truck passes by and you definitely feel the air move.
If we change the contour to pressure, and zoom in a bit, we can see areas that are affecting the drag on the bike. Anywhere that there is high pressure on the front side, and low pressure on the rear side is creating a pressure drag force. You can clearly see this on the front wheel. You can also see this on the headlight, which surprised me a bit. I was thinking the engine would have a lot of drag, but the headlights? Well it’s not just the headlights in this area; there are also the metal cross members that attach the risers to the body of the motorcycle. These are solid chunks of metal, so no air can get by these as well. Also, the headlights are experiencing fresh high speed air, so that’s creating more pressure. The engine is seeing the air that has been slowed when going around the front wheel, so the pressure isn’t as high.
Next I made an isosurface connecting places with the same velocity values, one at a speed of 45 mph, and another at 70 mph. I added some grid lines to help show the contours of the isosurface. This really helps to show the extent of the wake, with the 45 mph isosurface being the air that we have greatly reduced its speed, and 70 mph isosurface showing the extent of the air we have slightly disturbed. Now because the 70 mph isosurface extends so far, it’s hard to get a sense for it because it’s obscuring the chopper. So I made a cropped isosurface down the chopper centerline so we can see inside the isosurface.
One thing of interest to me was the aerodynamics of the wheels. I know in open wheel racing, the spent a lot of research on the aerodynamics of exposed wheels. When it comes to custom choppers, something that is always done is creating a cool looking wheel. I’m sure every design could be improved aerodynamically. I created a surface plot of velocity on the front wheel to see what I could see. Obviously velocity is zero on a surface, so this plot is offset into the air slightly to get the near wall air speed.
You can see in the tread area of the middle of the wheel there is a low speed zone. You can also see on the spokes near the rim, there is a lot of yellow high speed air, whereas near the hub behind the brake disk the air speed is pretty low. Now these wheels are pretty simple by chopper standards, so it would be interesting in the future to compare the drag and torque results for these wheels verses some other design.
Of course any analysis isn’t complete without some videos. First are some streamlines flowing around the bike. The interesting thing to me, is the air going by the front wheel. Even though the wheel is rotating clockwise, you can see how the air moving over top of the wheel flows downward, because of the engine. You can see some clockwise flow within the wheel, but in general you have competing airflows in this area. Now because the air has a downward trajectory as it goes towards the engine, I think this would reduce the drag effects of the engine because we aren’t getting any stagnation points on the front of the engine. The air is being deflected downward and away from the engine.
Another air flow we maybe interested in is seeing where the exhaust air goes. Every custom chopper design has to have cool pipes, so some designs may put the hot exhaust gases where we don’t want them. Looking at the animation, we can see some exhaust gas gets entrained with the air circulating around the rear wheel, but the exhaust gas cools very quickly. Again, I took an educated guess for the gas flow rate here.
Similarly, we would be interested in seeing how the air gets to our engine intake. We don’t want to put our intake in an area with poor airflow that could starve the engine and reduce the power output. We can see here, that other than a little wiggle to go around the front wheel, the air has a nice straight line into the intake.
Of course we could dissect this model further and looking forces and torques numbers on different components or the entire chopper. But I think I’ve learned a good amount about the aerodynamics of choppers. I hope you did to. This took very little time to analyze as I had the CAD geometry, and FloEFD works within the CAD system and automatically meshes the model. As such, I don’t want to get too deep into dissecting these results. Instead I would like to save some time to work on the BBQ analysis before the summer is over, so we will skip the numbers on this analysis. Until next time…
Downdraft Range Vent Hood Flow Analysis
One of the cool kitchen features that caught our attention when we saw what would become our house was that the stove had a retractable range vent. It would pop up behind the stove, and sucks the air away from the side, then down through the counter, down through the floor, and out the side of the wall. What really makes these things great is they open up the whole kitchen, as we have one big kitchen/dining room/living room. A traditional stove hood would have been right in the middle of the room, and would have covered up our large patio windows.
Now, of course after we moved in and started using it, I was less than impressed. The range has the largest gas burners at the front, so those are the ones we use the most for boiling pasta water or cooking on a skillet. Steam is great to see air flow, as it’s a white color, and I could see it wasn’t making its way into the vent (see link below for example downdraft vent with steam from pot). So of course, this was one of the household applications I wanted to analyze with FloEFD, to see what was going on.
First, I looked up our downdraft vent brand, and found the user/installation manual. It mentioned the blower was was rated to 600 CFM. Perfect, if I ignored filter and duct loses, I had everything I needed to model this ventilation system. Of course, being CAD lazy, I needed to find a model of a cook top. Luckily, I found one on grabcad.com. After building a pot of water to include in the analysis, things didn’t quite look right, the pot seemed to be too large for the stove. Seems the free stove model I got was from somewhere with smaller kitchens then here in California. I decided to press on and see what the results would tell me. Maybe downdraft vents are designed more for use in smaller kitchens? FloEFD would tell the story.
The last bit of info I needed for my analysis was the flow rate of the steam from the boiling pot, as I wasn’t going to actually analyze that process. As with any engineer, I have a boss who wants me to get my work done as fast as possible, and I need my computer to run simulations as fast as possible so I can have it simulate something else. It was much faster to provide a flow rate for steam. Looking into it, according to the internet, a gallon of water will create 1700 cubic feet of steam. According to my box of Mac and Cheese, I should boil 6 cups of water which is roughly a half a gallon. From experience, I know that amount will boil off in about 2 hours, give or take. So we have a steam flow rate of about 7 CFM, higher than you’d think right?
Now the fun part, let’s look at the results. I could have put a bunch of pots on the stove, as I’m sure the performance of the vent would be lower if our steam had obstacles in its way. But, I rarely have a bunch of things on the stove at once, so I thought that was a little unfair to the downdraft vent. I also thought of including the carbon dioxide and monoxide coming from the burner, which is very important as it’s the real reason we have the ventilation in the first place. But I believed that as the pot is already so high as to be in line with the vent, it would be the tougher task to suck into the vent. The exhaust gases from the burner are lower, so have more time to be pulled into the vent as they rise.
Below are some results looking at the steam concentration (isosurfaces and flow trajectories). You can see from the isosurfaces all the steam is going into the vent. I think using spaghetti shaped flow lines made sense for my boiling pasta water pot You can also see though, that a fair amount of the flow that makes it into the downdraft vent comes from behind the vent, not infront of it. So I believe this is a design issue unique to the downdraft vent. I don’t think traditional range hoods have an issue of sucking in air from near the ceiling, it’s all coming upward from the stove. Plus the hot gases/steam are already rising up because of bouyancy. Maybe some extendable side ducting fins would help the vent suck in 600 CFM from the front of the vent, the important stove area, instead of just the surroundings.
It seems like my vent is doing a perfect job. Now we get to something I’ve done a thousand times working on support cases. My simulation results don’t match my “experimental” results (i.e. cooking observations). This is where my analysis stalled for a bit while I thought long and hard for that one key fact that I was missing. It’s always a lot easier to find someone elses mistake then it is to find your own…
What I realized is that we never use the downdraft vent on high power. It sounds like a freight train, and completely drowns out any conversation which is the whole point of the open concept room, to allow you to cook in the kitchen and still talk with people in the living room at the same time. I always use the medium setting because while it’s still not whisper quiet, it can be talked over.
I had no idea what flow rate the medium setting would provide, so I set the flow rate to 300 CFM, half of the 600 CFM high setting. That felt like medium to me. Here are the FloEFD results.
Well now the experiment and simulation results are matching, which is always a good feeling. It seems like on medium setting, the vent is doing basically nothing to the steam leaving the pot. This was exactly what I was seeing when I was cooking, but I wasn’t sure if the vent was at least diluting the steam by inducing fresh air into the cooktop area. During my research for this, it seems like other models of downdraft vents have higher CFM’s. Maybe my vent was a a low end unit, and the blower is underpowered for this application? Maybe, but I think if the vent was able to better direct where it was getting the air from, the important stove air, instead of getting air from behind the vent, then even on the medium setting it would likely perform a lot better. I also think though if the unit was quieter, it would be a lot better. To have 600 CFM flowing through the narrow thickness of this vent, it’s going to be loud unless analysis is done to minimize these effects. Maybe that’s an analysis for another day. In the meantime, I’ll simply crank the volume on my TV when I cook so I can use the maximum setting on the vent.
Next blog topic is close to my heart. Analyzing my propane grill with my smoker box to fine hotspots/cold spots, and smoke distribution within the grilling area.
Hockey Puck Aerodynamics
This blog has been in the works for about a year. I think it was worth the wait. Let me explain, as a number of things were at play here. First, I had just attended a Siemens NX CAD system training in advance of our release of FloEFD for NX last year. So I was looking for an application I could cut my teeth on for my first FloEFD for NX analysis. Something with CAD requirements that were simple enough that I could create with my beginner NX CAD skills. Also, I had just done a presentation at PTC world on CFD in sports, basically going over some of our customers FloEFD analysis on their sporting equipment, like this company that analyzed their sail boats http://www.mentor.com/products/mechanical/success/wb-sails .
Now with this being one year ago, it was in the heart of hockey playoff season, and my team, the San Jose Sharks were battling with Vancouver in the conference finals for a spot in the Stanley cup. So I was struggling to figure out something related to hockey that I could do a CFD analysis on. Then one night I was throwing around the football, and remembered some show about the history of football and how the football was originally pretty round, but over time became enlongated to allow for accurate forward passes. Basically the shape of the football was determined by its aerodynamics. That made me think about hockey pucks. Hockey pucks can reach speeds of 110 mph on the fastest slap shots. The shape of the puck though has stayed constant for 100 years I bet. It’s flat surfaces are needed to keep the puck flat on the ice and not bounce during passes, but no thought has been given to it’s aerodynamics.
So I had my idea, to analyze the hockey puck. But then, the sharks lost in the conference finals, and really, I didn’t want to think about hockey after that heartbreaking defeat. It was my plan to wait till the San Jose Sharks made it to the cup this year for this blog, but unfortunately they lost in the first round to St Louis. Not wanting to wait another year, I decided to finish this off now.
My specific focus on this analysis was the grip pattern around the puck. I’m guessing it helps with keeping the puck on the stick, but what is the consequence on the drag on the puck? I mean, hockey players switched in droves to the composite hockey sticks for the better performance on their shots, so why wouldn’t the NHL be interested in better puck performance. Now, I found a NX CAD part where someone had made a knurled pattern on a shaft. I simply swished the shaft down to the dimensions of a puck, and adjusted the knurl pattern to the spacing on a puck. Now, as a Canadian, I have a hockey puck on my desk at all times, so I did see that the grip pattern is a bit different then the knurling pattern, in that knurling is cutting the cross hatching into the object, where the puck has the cross hatch pattern raised. My CAD skills were beginner, so I went with what I had.
Below are some images and animations of the results for the knurled hockey puck at 100 mph. I utilized FloEFD’s solution adaptive mesh, as I figured the drag results would be very sensitive to accurately capturing the low speed wake. You can see the mesh in the one image, and how the mesh flows the wake downstream. The main thing I wanted was the drag, which I found to be 1.18 N. Very small amount of force.
Then I analyzed with no knurling/smooth sides, to see whether the knurling is like the dimpling on a golf ball, helping the boundary layer reattach sooner and reduce the drag on the ball. Below are some images of the results. For the most part the results seem the same. You can see the wake though seems more a-symmetric (lower speed on one side). This indicates that now there is a oscillation in the shed vortex so this lower speed air would move back and forth. As this analysis was run steady state, this is just a snapshot in time, so the low speed area happened to be on that side when the analysis stopped. The isosurface to me, seems to be smaller (not as wide and tapers more downstream compared to previously), an indication of less drag. And sure enough, FloEFD says the drag is 0.9 N, again a small amount, but a full 23.7% improvement!
Now, to be a true aerodynamic characterization, I would need to solve at different angles of attack, and I’m not sure what kind of spin a hockey player imparts on his shot, but I’m sure that would influence the results. Still, I hope this opens the NHL’s eyes to looking into the important aspects aerodynamics play in their sport. 23.7% faster shots would make it harder on goalies and increase scoring, which is always a good thing.
Smoke, Smoke, Everywhere?
This has the possibility to be my most unpopular blog ever, which I’m ok with. The topic of this blog is smoking, specifically the second hand kind. Let me start by saying I am not a smoker. I don’t care for the smell, and am happy that it is not allowed in most building, Vegas casino’s being the main exception. My mom was for a long time though, and in general it seems to me that smokers have become the easy target. They are currently a voting minority so it’s easy to impose laws that take away some of their rights, which I think is wrong.
Let me clarify, in California, a lot of municipalities have 25 ft bans on smoking near doorways or windows. Ok, I can see that, but why 25 ft? Who came up with that distance? Did anyone study how smoke diffuses in the air and figure that after 25 ft the smoke has diffused to a concentration low enough to not cause cancer? I haven’t been able to find it. But ok, I understand that if I had an office window that happened to be by the local smokers spot, I would get annoyed by the daily smell of smoke, but would be forced to stay there so I could complete my job.
Then California in 2010 tried to ban smoking in State parks and beaches. Ok, I can understand that cigarettes may lead to forest fires and litter issues, but it was veto’ed by the Governor Schwarzenegger. Currently many municipalities in California have their own bans, like in San Jose, where smoking in public parks is completely banned. This is where I thought it would be good to analyze the problem in FloEFD, Mentor Graphics general purpose CFD software.
The purpose of this study is to see what type of 2nd hand smoke concentrations an innocent bystander might be expected to experience if someone nearby was smoking. First I needed a CAD model of the park, as FloEFD is a CAD embedded analysis software. Usually as a design engineer, the CAD is the easy part, as it has already been build. For me, that is my toughest bit as I usually don’t have CAD to start with. Being CAD lazy, I found a free CAD model of a skate park on www.grabcad.com. It isn’t the trees, grass and hills I was picturing when I started this, but it is a public park so I went with it. With that, I put in a couple CAD people, one to be the smoker, and 2 to be the potential second hand smokers.
Now I needed to come up with the analysis settings. I could find breathing rates and how much air a person breathes (~ 0.5 ft3/min), so I used that as boundary conditions on their nostrils. I assumed the smoker was exhaling 100% smoke, for every breath he took. This was highly unrealistic in my opinion, but as I couldn’t find a percentage number to put on this, I had to use a conservative number. Now, I wanted to capture any buoyancy effects, so I had the exhaled smoke breath coming out at 37 degC, while my outside air temperature was 20 degC. For the mesh setup, I didn’t know where the smoke would go, but knew i needed a finer mesh in the smoke cloud to get accurate concentration results. As such, I enabled FloEFD’s solution adaptive mesher, which will periodically pause the solve, go through the results, and add mesh where it’s needed. Very handy indeed.
Lastly, I modeled the wind. I also modeled a no wind condition, but then the hotter smoke just go upwards and never did get anywhere near my other park goers. See the figure below of an isosurface showing everywhere that the smoke is 1e-6 in concentration (1 part per million), and an animation of that smoke cloud vs time when he starts smoking and puffing. Basically the smoker is covered in a cloud of stinky smoke, likely why all cloths smell like smoke when you go to a Las Vegas casino or bar. It seems like being a couple feet away from him, you would be fairly safe, well maybe not your shoes, but still 25 feet away seems very arbitrary distance considering these results.
Looking at the streamlines, you can see the contrast in forces on the smoke. Here, it’s clear that because the smoke was exhaled through his nose, it has a lot of downward momentum before any buoyancy forces can take over and bring the smoke back upwards. Had I analyzed using the mouth as my boundary condition, the results might be different. But, it the grand scheme of things, how often is the air completely still for an extended period of time. Not that likely, so I didn’t bother running a mouth breathing case.
For my wind cases, I didn’t want to have a really stiff breeze, because all that fresh wind air would dilute my smoke, yet I wanted to have enough wind to push it towards my non-smokers. I wanted to model the worst case wind, so I chose ~3 m/s. If you read my crawlspace moisture blog, you know that 1 m/s is the bare minimum a person could feel, so 3 m/s seemed like a reasonably average breeze.
Below are my angled wind results, where I directed the wind to make the one bystander right in the smoker’s wake. You can see he is right in the smoke area, but the concentration is very small. Again, I haven’t seen at what concentration of smoke causes cancer (like they provide for mercury or any other carcinogen), so I thought 1e-6 seemed like a very small amount (1e-6 volume fraction of smoke = 1 ppm, so very low). Also, again, I assumed 100% of the exhaled breath was smoke, so this is the worst possible 1 ppm smoke cloud and in reality would be smaller.
Looking at a contour plot of smoke concentration from 100 ppm to 1 ppm, you can see while he is in the smoke zone, he is in the low concentration area. How low, well I tracked the air entering his nose, and the table below shows he was getting an average of 8.9 ppm of smoke. What I see though, is the smoke zone is about 6 feet wide. I would just move out of that area to the clean air if I smelled smoke. Our other non-smoker is completely clear of any smoke. Of course, if you were “dropping in” to that bowl or doing any “ally’s” on that rail, you would be in the eye of the smoke cloud.
On the other hand, if we look at a contour plot of smoke concentration from 1000 ppm to 0 ppm, we can see that in the “smoke cloud” the concentration stays relatively high. I was surprised by this, as our skate park has a lot of ramps and things to create turbulence, which I thought would mix in a lot of fresh air and drop the concentration levels. In fact, looking at my measurement of distance from the smoker’s nose to the non-smokers nose, we can see they are about 40 feet apart, and the concentration in the smoke cloud makes it well past the non-smoker. So a 25 foot ban on smoking near doors and windows seems a little too low after looking at this (again, this is worst case with 100% of exhaled air being smoke). The smoke cloud is a local effect, but it propagates a long way down stream.
To conclude, I don’t have any specific conclusions. I just saw an issue that was perfect for CFD to look into, and FloEFD helped me answer these questions in a minimum amount of time. As is usually the case when it comes to fluid dynamics, usually I have an idea of the answer at the beginning, but the results are rarely what I anticipated, even after years of doing this. In the end I think I showed both sides of the argument, smoke does travel a fair amount of distance, but it doesn’t blanket and poison the entire park. Should it be banned from all city parks and beaches? I don’t know if my results would swing the argument one way or the other. The X factor is the amount of smoke concentration that repeated exposer to would cause cancer, which medical science needs to answer first.
Next blog will be something lighter, aerodynamics of hockey pucks
The other day I was watching this show on the recent cruise ship sinking over in Italy. It was very interesting. They had ship experts looking over the data from the cruise ship, specifically GPS position and speed. One of the important factors that they said saved a lot of lives that night was the wind. After the ship had hit the underwater rocks and started taking on water, it continued out to deeper water. When the engines became submerged and stopped working, the ships momentum kept it moving out further from the island for a while. But, slowly the ship lost speed, and because there was an onshore wind of 25 miles per hour that night, it was first twisted/yawed to be perpendicular to the wind, then pushed back to shore.
The ship experts credit this as saving hundreds of lives that night, as people weren’t ordered to abandon ship for a long period of time. So long a time in fact that half the life boats couldn’t be used because the ship was leaning at too great an angle by the time people were told to abandon the ship. Many people ended up swimming for shore, which would not have been possible had it not been for the wind.
This made me wonder how much force it would take to move such a giant ship? I mean, this ship weighed 114,500 gross tons (256,480,000 lbs), and the wind was able to blow it a large distance in about an hour or 2? I needed to see what amount of force the air could exert on a ship this big.
I went on to grabcad.com, and found a cruise ship model. It was more for a desktop type model, but I easily scaled up the CAD to the dimensions of the real ship. Then, in FloEFD I simply defined the 25 mile/hour wind, set some goals to track the force on the ship, and the roll torque. Then I started to solve. It really was a very easy model to setup.
Here is what I saw. In this first image you get an idea of how large the wake is for a ship this big. I also displayed the mesh so you could see how it automatically adapted to the ship, and also the wake, which is very important to getting accurate results.
Zooming in, we get more detail on the mesh local to the ship. You will also notice an area of light blue higher speed air coming out of the middle portion of the ship.
What is happening here, is since I didn’t make this CAD myself, I missed the fact that there were no windows in this ship CAD, just holes through the ship. You can see in this image here the air moving through the ship where cabins should be. Obviously this isn’t correct, so a corrected model was run with the holes sealed.
I also plotted velocity vectors with the velocity contours to see how the air was moving as it came toward the ship.
- Cruise Ship Wind Wake with Velocity Vectors
- From the top view, we can see the width of the wake downstream. Also, the mesh was plotted so we could see how the mesh adapts to adding cells in the areas of velocity changes to accurately capture the wake structure. The adaptive mesh is one of my favorite FloEFD features, as it saves so much time getting an accurate mesh.
We can get a better feel for the airflow using streamline animations, like the 3 below showing different views of the same streamlines.
Another good way of getting a feel for the extent of the low speed air is an isosurface, which is a 3D plot that connects points with the same value to make a surface. Here I’m looking at anywhere that the airspeed is 15 miles/hr, a good amount slower than the 25 miles/hr wind. You can see the extent of the air being slowed down, which gives an indication to the forces affecting the ship.
Similarly a surface pressure plot will show where the high pressure regions, which when added up over the entire surface provide the force on that object.
But of course what we really want is the actual values for the force and the torques. FloEFD can output this into excel spreadsheet tables. Below are the results from the second analysis with the cabin holes “sealed”.
My main surprise is how low the actual force was compared to the weight of the ship. The ship weighs 256,480,000 lbs! 264,175 lbs of force is about 0.1% of the ship weight, yet it was enough to move the ship back to shore. Granted, we didn’t include any ocean current/wave effects in this analysis, which may have played a part, but as there was no data for that, I couldn’t include this into the simulation.
The more I think about it though, the more I can see that the force shouldn’t be massive. If it was, imagine how many docks would be crushed when a stiff wind blew in on a ship that was docked. Also, unlike most things that we can reference when thinking about forces, a ship isn’t on solid ground. There are no ground forces to resist this wind force. Similar to those “World’s Strongest Man” competitions where men pull airplanes or 18 wheel transport trucks, the pulling force is small compared to the size of the object being pulled, which is possible because the tires on the objects. Or how little tugboats can move ships much larger then they are.
In the end, what I take out of this analysis was that I was really surprised at the size of the wind wake behind the ship at about 1450 ft. But more then that, I was amazed at the tiny amount of wind force that moved this massive boat to shore and saved so many lives that day.
It’s that time of the year, when the weather turns cold and people start to think about winterizing their home to reduce heating costs. Usually it takes the first winter heating bill to provide the motivation to undertake this task. With this in mind, I would like to talk about pipe insulation. Specifically, the foam wrap insulation you can find at any hardware store (http://www.homedepot.com/h_d1/N-5yc1v/R-202318552/h_d2/ProductDisplay?langId=-1&storeId=10051&catalogId=10053)
In my case, I started to look into pipe insulation for an entirely different reason. During our home inspection, we were told that much of our copper pipe was run alongside the HVAC ductwork. In some spots, the metal strapping for the duct was touching the copper, which would cause some galvanic corrosion. The simple fix would be to wrap the pipe with something to prevent this contact. When I stumbled across this foam pipe insulation for about a buck for 6 feet, I was sure I had found my answer.
Since I was going to the trouble of going into the crawlspace, I figured I should buy a bunch of these and wrap as much copper pipe as possible, because my pipe is exposed to the cold outside air in the crawlspace, so I thought there could be quite a bit of energy efficiency gained. Plus, as the master bathroom is at the opposite end of the house, I had noticed it takes a minute or two in the morning to get hot water. My new hope was to be able to shower or wash my hands in the morning without having to waste water and time waiting for it to get hot.
During the installation of these foam pipe covers, I found it difficult to get the foam insulation in different areas, whether because of T-joints, strapping or bends. I also didn’t buy enough foam, so there was still some exposed copper. This is when I started to think about running a CFD analysis on this problem. If there is some exposed copper, because of its excellent conductivity, will the heat just move to this opening, rendering all my insulation efforts moot? Did I need to make the long crawl through my crawlspace to put more pipe wrap on, or were these little portions of exposed pipe inconsequential compared to the many feet of newly insulated pipe.
For this analysis, I used our general purpose CFD tool, FloEFD. I needed some baseline numbers, so I modeled a 1 meter length of copper pipe, then I would analyze that copper pipe completely covered with the insulation (best case), then introduce a representative “gap” in that insulation for my current setup. From some research I found that hot water comes from the tank at about 50 degC, and a shower can draw about 2.5 gal/min. For the air, I wanted to simulate the worst case air temperature, which I think in the winter in my crawl space would likely be about 5 degC (above freezing for sure, though preventing pipe freezing is an added benefit of these pipe insulations).
Now we all know convection heat transfer improves with air velocity, so I wasn’t sure where to go here. From my previous crawlspace blog, where I looked at my soil water issue, I found the air speed down there was sluggish to say the least. Yet, the copper pipes running to my shower run within 1 ft of the crawl space vents, and I could feel a breeze at that location. So I decided I would need to run a no wind and a 1 m/s wind case. Below are my results.
The other main result was that the heat transfer rate was basically the same. In fact it’s slightly worse with insulation compared to without, due to the insulation having a larger surface area.
Now, looking at the results for the with wind case, where we see a bit of a reversal on the heat loss trend. Now that forced convection is dominant, the increase in surface area for the insulated pipe doesn’t seem to be a factor. At the end of the day though, the water temperature is still pretty much the same value.
It’s at this point when I started to look at this problem in a different light, as I didn’t want to have wasted money putting on insulation that isn’t effective. My new thinking is that a steady state analysis of this problem is not ideal. We will never be running the water for hours on end. My goal is to have the hot water that is in the pipe to stay hot for as long as possible so that it doesn’t take 5 minutes of running the tap to get hot water at the sink/shower. This wastes water and energy and my money, and that’s what I’m hoping gives me my ROI for my foam insulation investment.
With that, I decided I needed to simulate this as a transient analysis, starting from when the water has stopped running and timing how long it takes for the water to sufficiently cool. I figure that would be somewhere around room temperature when you would think the water isn’t “hot”.
At this point I only simulated the wind case and full/no foam, as I’m more interested in the worst case scenario (and justifying my insulation purchase).
Now this is what I’m talking about. Without insulation, the water in that pipe gets to a chilly 10 degC in about 11-12 minutes, whereas with insulation, the water doesn’t get that cold till 100 minutes. That’s a 9 times improvement for $1 per 6 feet of pipe. It’s hard to argue with that return on investment. Now, I usually shower first thing in the morning, so the water won’t stay hot throughout the night, so I’m out of luck there. For everyday washing of hands and kitchen stuff, we will definitely be wasting less water because of these pipe wraps, and that’s what matters.
About Travis Mikjaniec's Blog
- Data Center Presentation at ASME in Texas
- Mentor Graphics Data Center Design using CFD Modeling
- CFD and the Olympics
- Chopper CFD using FloEFD
- Downdraft Vent Hood Flow Analysis
- Hockey Puck Aerodynamics
- January 2013 (1)
- December 2012 (1)
- August 2012 (1)
- July 2012 (1)
- June 2012 (1)
- May 2012 (2)
- March 2012 (1)
- January 2012 (1)