Why do materials fail at different rates?

[Moon Bee]

Why do people turn on several computers then leave them there ?
I think one/4-5 comps

-Moon Honey Bee

 

[Cyrus]

Computer that turns itself off.

 

[Mallignamius]

I leave mine on 24/7 to avoid thermal shock.

 

[dontdisturbmycircles]

Computer that turns itself off.

lol, right when I saw the Top Gear guy I knew it was going to be funny.

I don't always turn my computer off because I am downloading something or whatever. Or I have something running that I would not like to stop. But its also more nice to not have to wait while it boots up when I come back.

If someone had 4 computers turned on and they were the only one there... well, they are probably working with them. I can't think of a reason to have 4 computers unless you run a small computer lab.

 

[Danger]

I leave mine on 24/7 to avoid thermal shock.

Ditto that. I don't even shut off this iBook, but it will do it on its own if left for more than a week or so (LIon battery, which will be destroyed if it gets completely drained).

 

[russ_watters]

Thermal shock? Are you guys serious? That simply isn't an issue. Thermal shock is what happens when you get a rapid temperature change, like the cracking of an ice cube in a glass of water. Besides most of the components of your computer not being all that sensitive to it (you can quench red-hot metal and it won't break...), taking 30 seconds to warm up does not consitute "thermal shock".

I tend to leave mine on in the winter because it doesn't hurt my electric bill any to have it on, but I shut them off in the summer when I'm not using them because it costs a lot to leave them on.

 

[Mallignamius]

I lost a computer last winter to it. It's specifically my circumstances here where I don't bother with the furnace much. 50° F and I'm happy.

But when I plugged mine in one cold winter day, it booted up on its own. Poof! No more computer. I had intended to wait for room temp to go up before booting.

 

[Cyrus]

Thermal shock? That simply isn't an issue. Thermal shock is what happens when you get a rapid temperature change, like the cracking of an ice cube in a glass of water. Besides most of the components of your computer not being all that sensitive to it (you can quench red-hot metal and it won't break...), taking 30 seconds to warm up does not consitute "thermal shock".

Thermal shock does not mean rapid. You get thermal shock when you have a large temperature gradient and associated mechanical stresses. Computer chips are made of ceramics, not metals, and are VERY sensitive to thermal shock if they overheat. Anyone whose burned out a chip knows, they will crack when they go 'poof'. Most silicons fail for a gradient of 300C.

Saying its rapid is rather vague. Rapid is different from one material to another. I can slowly heat a computer chip so that one surface is at 0C and the other is at 300C. It will fail from thermal stresses, but its not rapid.

See: http://www.hexoloy.com/hexoloy_data-sheets/technical-data/1006-6.pdf/attachment_view/file

None of these equations have time as a variable. They all deal with spatial temperature distributions and mechanical properties (poisson's ratio, youngs modulus).

 

[DaveC426913]

Why do people turn on several computers then leave them there ?

1] Computers are fairly low energy consumers these days. It's not hard on the bill.
2] Most of the wear and tear on a computer occurs in booting up. It is easier on the computer to leave it running than to be constantly turning it on and off.

 

[JasonRox]

I've left my laptop out in my car in cold for a whole day and turned it out as soon as I brought it without any problems.

 

[robphy]

I keep mine on 24/7 to act as a fileserver or as computation server... I'd turn it off more if I can get WakeOnLAN to work reliably with my setup.

Many computers now have power saving modes...
at the CPU level, features like CPU throttling and better heat management;
at the system level, Standby and Hibernate, as well as ,... and hopefully soon, "Instant On" [e.g., resume from hibernating to solid state drives]. Often, I am impatient to wait for it to boot up.

 

[Danger]

and hopefully soon, "Instant On" [e.g., resume from hibernating to solid state drives].

Maybe that's another advantage of a Mac. If this thing is just resting, touching anything or opening the clamshell wakes it up. If it's hibernating, hitting the 'shift' key does it. Only when it has shut itself off do I need to restart it. Since I don't leave it alone long enough for that to happen, it's a minor issue.
I didn't realize that 'Windows-burners' don't do that.

 

[robphy]

Maybe that's another advantage of a Mac. If this thing is just resting, touching anything or opening the clamshell wakes it up. If it's hibernating, hitting the 'shift' key does it. Only when it has shut itself off do I need to restart it. Since I don't leave it alone long enough for that to happen, it's a minor issue.
I didn't realize that 'Windows-burners' don't do that.

My Dell laptop running XP can (on demand [say, by closing the cover] or after a preset time) standby [which can be waken by the keyboard] and hibernate (i.e. ram to disk, then power off) [which can be resumed by pushing the on/off button].

Did Macs always have the features you describe? Or is this a feature of the newer now-Intel-based Macs?

 

[Astronuc]

Thermal shock does not mean rapid. You get thermal shock when you have a large temperature gradient and associated mechanical stresses.

Umm, Russ is right. The term 'shock' normally implies a transient phenomenon, which is usually 'rapid'.

Certainly thermal gradients produced thermally induced stresses.

Microchips are mostly semiconductors with metal conductors insulated with metal oxides (ceramics). Each material has a different coefficient of thermal expansion, although the metals and semi-conductor have similar CTE's as compared to ceramics, IIRC.

Thermal fatigue and wear and tear on HD's are a reason to leave them on. Similar experience to that of Mallignamius, I started a computer cold one time (I stupidly left in the car on a cold night), and the HD got damaged.

But leaving a computer on does not mean the monitor has to be on. Most PC's have a power reduction or standby mode.

 

[Moonbear]

Did Macs always have the features you describe? Or is this a feature of the newer now-Intel-based Macs?

Yes, those features have been around a long time on Macs. I didn't realize they didn't exist on PCs. Surely PCs are the same way, aren't they? Isn't that the whole point of a sleep mode, to be able to quickly wake it back up? If you had to wait for a full re-boot, you might as well turn it off.

 

[russ_watters]

To reconcile these:

Umm, Russ is right. The term 'shock' normally implies a transient phenomenon, which is usually 'rapid'.

None of these equations have time as a variable.

Yes, they do: time is the key part of heat transfer rate, which is where the large gradient comes from.

Thermal shock does not mean rapid. You get thermal shock when you have a large temperature gradient and associated mechanical stresses.

The theory presented by others was that the thermal shock happens at startup, where the gradient is caused by the rapid heating of a cold chip - exactly the situation of an ice cube in water (if they were correct about it being a large transient gradient...).

Thermal fatigue and wear and tear on HD's are a reason to leave them on. Similar experience to that of Mallignamius, I started a computer cold one time (I stupidly left in the car on a cold night), and the HD got damaged.

Now that is a bad idea - I've bought components in winter before and I have to resist even taking them out of the package before they warm up. Besides the moving components issue with hard drives, in winter you have a condensation issue too. I think condensation from going straight from the cold outside (using my telescope at night, in winter) to a humidified bedroom multiple times contributed to the death of my last laptop.

 

[russ_watters]

I lost a computer last winter to it. It's specifically my circumstances here where I don't bother with the furnace much. 50° F and I'm happy.

But when I plugged mine in one cold winter day, it booted up on its own. Poof! No more computer. I had intended to wait for room temp to go up before booting.

There are plenty of people out there who overlcock the crap out of their computers and use either water or phase-change cooling. The thermal gradients and heat transfer rate, and associated shock are much higher than what you would have seen.

But when I plugged mine in one cold winter day, it booted up on its own. Poof! No more computer.

That tells me you lost your computer to static electricity, which is a big problem in winter when the humidity is low. When you plugged-it in, the static discharge was enough to jump the on/off switch and fry the motherboard at the same time.

I've lost a couple of motherboards and a couple of USB flash drives this way - the computers we have at my office have badly grounded front USB ports.

 

[robphy]

Yes, those features have been around a long time on Macs. I didn't realize they didn't exist on PCs. Surely PCs are the same way, aren't they? Isn't that the whole point of a sleep mode, to be able to quickly wake it back up? If you had to wait for a full re-boot, you might as well turn it off.

I never said that features like standby and hibernate didn't exist on PCs.
My earliest statements are implicitly making reference to very old PCs.
The first time I used hibernate was with Windows 2000. Windows 98 already had that feature http://www.microsoft.com/windows98/usingwindows/work/articles/908Aug/hibernation.asp.
By the way, in my earlier reference to "instant on", I mean that the entering or resuming from hibernation [where the main power is now off] is very fast because the drive being used to store the ram is not your standard laptop hard-drive... but something like flash memory. Do any laptops (Mac or PC) have this feature yet?

 

[russ_watters]

Well, standby keeps your ram powered, which is why it is faster than hibernating. Flash standby would provide the speed of normal standby with the lower power consumption of hibernating.

 

[russ_watters]

1] Computers are fairly low energy consumers these days. It's not hard on the bill.

To be more specific, a 200w desktop computer costs about $20 a month to run (not including HVAC issues). People can decide or themselves if that is worth doing...

Laptops average a fraction of that - maybe 50 watts.

2] Most of the wear and tear on a computer occurs in booting up. It is easier on the computer to leave it running than to be constantly turning it on and off.

While technically true, that really isn't a big issue. I've never had a computer die due to wear and tear - only those who might keep one for 10+ years could run into that.

For example, if by having your computer off 90% of the time you reduce its lifespan by 90%, a 100,000 hour mtbf hard drive will still last on average, 11 years.

Interesting article on failure rates: http://www.netscape.com/viewstory/2...Highly+Exaggerated/article6404.htm&frame=true

 

[Moonbear]

I never said that features like standby and hibernate didn't exist on PCs.
My earliest statements are implicitly making reference to very old PCs.
The first time I used hibernate was with Windows 2000. Windows 98 already had that feature http://www.microsoft.com/windows98/usingwindows/work/articles/908Aug/hibernation.asp.

Okay, I misunderstood your point then. It happens. Macs probably got that feature about the same time as PCs then. The first time I remember using it was when those horrid fruit-flavor-colored iMacs came out which was somewhat contemporaneous with the Win 98 to 2000 switch. The Power Macs out just before that might have had the feature too, but I wasn't regularly using Macs before then so am not totally sure. I couldn't recall if Win 98 had a hard-drive sleep feature, or just a monitor display sleep. I also can't recall if the Macs before the iMac had anything other than a display sleep.

 

[Moonbear]

I tend to leave mine on in the winter because it doesn't hurt my electric bill any to have it on, but I shut them off in the summer when I'm not using them because it costs a lot to leave them on.

Can you explain that comment further? Why would the cost be different between summer and winter? Or is it that you keep the indoor temperatures different so the fan is running more in summer heat than in a cool house in winter?

 

[Cyrus]

Yes, they do: time is the key part of heat transfer rate, which is where the large gradient comes from.

Umm, Russ is right. The term 'shock' normally implies a transient phenomenon, which is usually 'rapid'

Im not sure about this. Temperature gradients don't come from transfer rates, but from material properties like heat transefer coefficient and biot number. Time has nothing to do with gradients.

Can you guys please show me how time is a factor? I'm not seeing it.

Take the rate equation:

q=-k(dT/dx)

time is not a variable in this equation, anywhere. The rate is a function of spatial and mechanical properties only.

 

[Astronuc]

Take the rate equation:

q=-k(dT/dx)

time is not a variable in this equation, anywhere. The rate is a function of spatial and mechanical properties only.

What are the units of k? W/m-K, and W = J/s.

The equation is for 'steady-state' heat conduction. But look at the time dependent heat conduction equation, and it's quite different.

Heat up and cool down are time varying situations. Thermal gradients will change - e.g. compare a mass of uniform temperature to one which has a thermal gradient - it takes time, and the heat has to be conducted.

The other factor regarding hot vs cold is the tolerances/clearances on the dimensions/parts, e.g. the arms which hold the transducers on an HD. HD's are manufactured at room temp - ~ 25°C. What happens when the HD is at 0°C - what is the contraction of the HD arm between 0 and 25°C.

It is rather interesting to see the variation in high precision instruments where the tolerances are measured in microns.

In mechanical systems where materials go from hot to cold, the idea is to heat up a system whereby the localized mechanical stresses can dissipate such that the local stress concentration never exceeds the threshold which promotes crack propagation (which usually assumes a pre-existing or nucleate flaw or imperfection).

 

[Cyrus]

What are the units of k? W/m-K W = J/s.

But k is a property of the material and is (more or less) constant for a limited temperature range. For this reason, the rate of heat transfer, q, is not a function of time, but a function of the material (k), and the spatial gradient (dT/dx).

 

[Cyrus]

Ah, yeah...here we go. Time is a factor in the temperature distribution for semi-infinite solids.

I was thinking of lumped capacitance which ignores spatial distributions within the material.

Still though, these equations http://www.hexoloy.com/hexoloy_data-sheets/technical-data/1006-6.pdf/attachment_view/file depend on the temperature difference $$\Delta T$$, no? It does not care how long it takes to get to some temperature difference. What matters is that you have a maximum allowable temperature difference.

 

[Astronuc]

But k is a property of the material and is (more or less) constant for a limited temperature range. For this reason, the rate of heat transfer, q, is not a function of time, but a function of the material (k), and the spatial gradient (dT/dx).

True - for steady-state.

But,

$$\rho\,{c_p}\,\frac{\partial{T}}{\partial{t}}\,-\,\nabla\cdot(k\nabla{T})\,=\,f$$

see equation 6 of this paper - http://www.ippt.gov.pl/~tzielins/Lectures/HeatPDE.paper.pdf

or see page 2 (Step 1) of this document
http://instruct1.cit.cornell.edu/courses/bee453/tut1comsol/Tutorial1.doc

 

[Astronuc]

Ah, yeah...here we go. Time is a factor in the temperature distribution for semi-infinite solids.

I was thinking of lumped capacitance which ignores spatial distributions within the material.

Still though, these equations http://www.hexoloy.com/hexoloy_data-sheets/technical-data/1006-6.pdf/attachment_view/file depend on the temperature difference $$\Delta T$$, no? It does not care how long it takes to get to some temperature difference. What matters is that you have a maximum allowable temperature difference.

Yes, $$\Delta T$$ is a key factor.

 

[Danger]

I'm not sure that what you guys are talking about is the same reason that I leave mine on. It's more of an erosion effect at electrical contact points that I was cautioned about years ago, as caused by an arc on start-up.

I also can't recall if the Macs before the iMac had anything other than a display sleep.

My LCII running OS 7 has the same stuff, and it's so old it has dinosaur **** on it.

 

[Cyrus]

Yes, $$\Delta T$$ is a key factor.

Would this not then imply that it does not have to be a rapid change?

 

[Astronuc]

Would this not then imply that it does not have to be a rapid change?

Certainly, not in all cases, but the term 'shock' usually infers a 'rapid' change in which a system cannot adapt readily, or dissipate the energy before failure occurs.

I could really use some graphics here. And I can't find an appropriate example online - assuming one exists. There are lots of journal articles one has to buy though. :grumpy:

 

[ShawnD]

I keep the computers running to prevent hardware failures associated with that shock given to all components when you turn the power on and off.

Does anybody remember that episode of mythbusters where they test whether or not leaving the lights on saves power? They showed that the saving power argument was bunk (obviously), but they also showed that when they turned the lights on and off at regular intervals not a single one of their lights lasted more than 2 months. When you're talking about a 60 cent light bulb, yeah I can see how saving power at the expense of the bulb is a good idea. When you start talking about a 2TB array made from 6 hard drives with a total cost of over $1000, does it really make sense to turn it on and off because you might save $5 over the course of a year?


When Russ said he leaves them on in winter, I think he means that power is never really wasted in winter time. Most of the electricity from the computer and monitor turns into heat. Same with light bulbs, the stove, the oven, etc. You're heating your home anyway, so "wasted" electricity in the form of heat isn't exactly wasteful. In summer it's the exact opposite. You're trying to cool your home, so any electric heat from your computer or stove or whatever is adding to the problem, and you end up using even more electricity to power fans and air conditioners.

 

[Cyrus]

Certainly, not in all cases, but the term 'shock' usually infers a 'rapid' change in which a system cannot adapt readily, or dissipate the energy before failure occurs.

I could really use some graphics here. And I can't find an appropriate example online - assuming one exists. There are lots of journal articles one has to buy though. :grumpy:

What I am getting at is this: Let's say, for example, that we have a semi-infinite solid. Then the temperature at any location is a function of:

T=T(x,h,a,k,t)

This means that the temperature at a given point x will vary with t only (considering h,a,k to be relatively constant over the temperature change for a given material).

So the 'speed' at which a material reaches failure is determined primarily by T=T(h,a,k) at a given location. One material will 'fail' faster than another material due to its physical properties (h,a,k), because x is held constant and time is allowed to increase the same for both samples. Time is simply telling you which one fails first. It does not say it failed because it was necessarily rapid.