I am interested in what values other engineers use or resources they pull from for calculating an 'average' heat load for a desktop computer. I would have replied to this post: thread403-217429, but it was closed a while ago, plus I feel like the information and resources there are now out of date (~14 years old at this point). I usually just use a standard value of 75 W. Thanks!
Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Heat Gain for Computers 4
- Thread starter IanVG
- Start date
-
1
- #2
EnergyProfessional
Mechanical
Usually as part of plug load 1W/ft2 unless there is a reason to have more. That would include printers, monitors, computers etc.
note there is a trend to use dual-monitors (I have two 43") and more powerful PCs unless we talk about simple admin work. Ultimately you almost to see how large cubicles are. The smaller, the more people (and monitors) per 10000 ft². Best to look at what the office layout and potential future expansion will look like. where i work, they squeeze more and more people into the same space (we now have 7x7 cubicles) and it started out at twice that size. So now in March we have 78°F in the office! so my 1W/ft² may not be enough.
note there is a trend to use dual-monitors (I have two 43") and more powerful PCs unless we talk about simple admin work. Ultimately you almost to see how large cubicles are. The smaller, the more people (and monitors) per 10000 ft². Best to look at what the office layout and potential future expansion will look like. where i work, they squeeze more and more people into the same space (we now have 7x7 cubicles) and it started out at twice that size. So now in March we have 78°F in the office! so my 1W/ft² may not be enough.
Note, however, dual monitors these days are mostly either LED or OLED illuminated, and dissipate way less power than the CRT monitors that are enshrined in some of the older documents. While there might be some tendency to power hogging in PCs, the modern laptop doesn't really dissipate that much more power than the ancient ones; there's been a de factor standard of keeping CPU power below either 45 W or 25 W. Certainly, my laptop AC adapter has steadily decreased in size since my first Dell, which had a monster adapter than weighed probably 3 pounds, compared with the latest one that weighs less than one pound. This is commensurate with battery life still going on an upward direction, or at least, staying at par.
As a general rule in my company, lower-level employees are situated two per 10x10 cubicle, while more senior employees are in same 10x10 cubicle by themselves or in walled offices that are more like 15x12.
TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
As a general rule in my company, lower-level employees are situated two per 10x10 cubicle, while more senior employees are in same 10x10 cubicle by themselves or in walled offices that are more like 15x12.
TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! faq731-376 forum1529 Entire Forum list
EnergyProfessional
Mechanical
I would use the person density that is assumed for ventilation unless it will be denser (but then you also have to up the ventilation).
Yes, new PCs and monitors are more efficient. But now we have 2 large LED monitors vs. one small CRT monitor. We also use 8+-core CPUs with powerful graphics cards. Laptops aren't really powerful for productivity work, like BIM. So unless you know this office just uses Excel and Word, I wouldn't assume a low power PC. Some people also end up with a desktop and a tablet (although not used simultaneously). The use of the office also may change during the life of the HVAC system and who knows how many big monitors etc. we will use in 10 years. There probably is a limit to how many people they can squeeze in an office and my 7x7 is as small as it gets. But I also saw 6x6 cubicles. Next office will look like a chicken farm. That obviously will increase power-density.
Most CPUs for "real" desktops are rated 65-90W TDP. and that is at base speed. It goes way above when they increase clock-speed and typically are limited by cooling. but the new Intel 8-core CPUs easily go over 200W. And that is just the CPU.... you also have a motherboard, power supply, RAM, SSD etc. Desktops typically don't run at full speed all the time, but I would assume at least 100W continuously. Then the 16-core PCs.... There is a reason the smallest brand name power supplies are rated to 400+W on the DC side (more on AC, which matters for cooling load). and 400W PSU is for low level PCs.
Yes, new PCs and monitors are more efficient. But now we have 2 large LED monitors vs. one small CRT monitor. We also use 8+-core CPUs with powerful graphics cards. Laptops aren't really powerful for productivity work, like BIM. So unless you know this office just uses Excel and Word, I wouldn't assume a low power PC. Some people also end up with a desktop and a tablet (although not used simultaneously). The use of the office also may change during the life of the HVAC system and who knows how many big monitors etc. we will use in 10 years. There probably is a limit to how many people they can squeeze in an office and my 7x7 is as small as it gets. But I also saw 6x6 cubicles. Next office will look like a chicken farm. That obviously will increase power-density.
Most CPUs for "real" desktops are rated 65-90W TDP. and that is at base speed. It goes way above when they increase clock-speed and typically are limited by cooling. but the new Intel 8-core CPUs easily go over 200W. And that is just the CPU.... you also have a motherboard, power supply, RAM, SSD etc. Desktops typically don't run at full speed all the time, but I would assume at least 100W continuously. Then the 16-core PCs.... There is a reason the smallest brand name power supplies are rated to 400+W on the DC side (more on AC, which matters for cooling load). and 400W PSU is for low level PCs.
John_Hanley
Computer
The simplest method is to read the power consumed on the UPS display. If you do not have a UPS, get one. Otherwise buy an inexpensive power usage meter.
I use a rule of thumb for modern (2020+) desktop computers. Typical i5/i7, two monitors: 250W or 900 BTUs. Now add another 50-100W or 170-340 BTUs for misc equipment such as iPads, LED desk lamps, printers, chargers, etc. Those numbers do not sound like much, but a room full can almost heat an office in the winter here in Seattle and make summers uncomfortable unless planned for in the HVAC design.
Powerful desktops with multi-core i7/i9 processors and large GPUs can easily hit 1,000W or 3,400 BTUs. Those computers will make a cubicle very warm.
I use a rule of thumb for modern (2020+) desktop computers. Typical i5/i7, two monitors: 250W or 900 BTUs. Now add another 50-100W or 170-340 BTUs for misc equipment such as iPads, LED desk lamps, printers, chargers, etc. Those numbers do not sound like much, but a room full can almost heat an office in the winter here in Seattle and make summers uncomfortable unless planned for in the HVAC design.
Powerful desktops with multi-core i7/i9 processors and large GPUs can easily hit 1,000W or 3,400 BTUs. Those computers will make a cubicle very warm.
-
1
- #7
EnergyProfessional
Mechanical
When you design HVAC for an office, you don't have information about the exact equipment. and the equipment known may not be in use in 3 years. so you need to use a design value that will account for future equipment (to a degree).
Nameplate power also isn't useful since that is the design power, not typical power use. And a PC doesn't have a nameplate that shows power use. Best you can get is the PSU data, but that is much higher than the components installed.
Some manufacturers give you the electrical data, and a separate cooling load. but this is more for servers, switches etc. but gain, you also don't design a server room or switch closet for specific equipment that may be obsolete in a year, but you use design values.
Nameplate power also isn't useful since that is the design power, not typical power use. And a PC doesn't have a nameplate that shows power use. Best you can get is the PSU data, but that is much higher than the components installed.
Some manufacturers give you the electrical data, and a separate cooling load. but this is more for servers, switches etc. but gain, you also don't design a server room or switch closet for specific equipment that may be obsolete in a year, but you use design values.
-
1
- #8
John_Hanley
Computer
I have designed data centers and computer-based offices many times over the decades. I am a Google architect. For years the trend line was lower power consumption each year. However, with the current trend of more CPU power and power-hungry GPUs, that trend is no longer as true. As AI and ML begin being deployed everywhere, there will be an increased power consumption trend which might go back down in five or ten years with increased CPU/GPU efficiencies. For now, the best is to budget a per person compute workload and adjust over time as trends change.
This budget is not as simple as reading the spec sheet or nameplate. CPUs are variable speed, have low power modes, sleep modes, and full-on computing high power modes. Just like we must calculate heating and cooling BTUs by using numbers, heating degree days, R-values, etc, and not guesses, we must also actually measure what a computer uses over an extended period of time and during peak power consumption during office working hours. During nights and weekends, most systems should be configured to go to sleep reducing power consumption. It is during the working hours when people are also present that the loads must be managed for people's comfort.
Note: when we design an office or data center, we do specify the estimated power consumption of all equipment. In high-density workplaces, this is an important item to determine, for example, the minimum CFM for HVAC cooling. Assuming that equipment is ordered every thirty months in volume, the power requirements can be factored into purchase decisions. That process can provide a feedback loop to the mechanical engineering department so that they can forecast future HVAC requirements. The days of over-provisioning must go away as net-zero buildings become a reality and not a dream.
Now some will say, "but we cannot estimate or forecast what equipment will be used." In some rare cases that might be true. However, a budget would be designed and then that budget can be incorporated into future planning or, for example, a lease or purchase agreement. Just like power plants have peaker plants, modern buildings can also have standby equipment that can absorb peak and/or unexpected loads.
This budget is not as simple as reading the spec sheet or nameplate. CPUs are variable speed, have low power modes, sleep modes, and full-on computing high power modes. Just like we must calculate heating and cooling BTUs by using numbers, heating degree days, R-values, etc, and not guesses, we must also actually measure what a computer uses over an extended period of time and during peak power consumption during office working hours. During nights and weekends, most systems should be configured to go to sleep reducing power consumption. It is during the working hours when people are also present that the loads must be managed for people's comfort.
Note: when we design an office or data center, we do specify the estimated power consumption of all equipment. In high-density workplaces, this is an important item to determine, for example, the minimum CFM for HVAC cooling. Assuming that equipment is ordered every thirty months in volume, the power requirements can be factored into purchase decisions. That process can provide a feedback loop to the mechanical engineering department so that they can forecast future HVAC requirements. The days of over-provisioning must go away as net-zero buildings become a reality and not a dream.
Now some will say, "but we cannot estimate or forecast what equipment will be used." In some rare cases that might be true. However, a budget would be designed and then that budget can be incorporated into future planning or, for example, a lease or purchase agreement. Just like power plants have peaker plants, modern buildings can also have standby equipment that can absorb peak and/or unexpected loads.
-
1
- #9
- Thread starter
- #10
Thanks everyone for the great information. I kind of expected that the best solution would be observation and application of data gleaned from that observation to future cases. Unfortunately, I do not (yet!) have access to that kind of data. I work for a large university (~500 buildings) and for the time being, a significant workload of mine is renovating classroom/research/office/lobby/conference room spaces (usually from one space use to another or sometimes just a facelift). On my site visits I try to get as much information as possible from the customer (usually the dean/facility manager/admin personnel of some sort), but it does not always line with reality (I let them know that their comfort is also dependent to some degree on their ability to forecast # of people and equipment used in that space). In any case, no one so far has been able to get me an equipment list for the kinds of computers (depending on department IT is budget-constrained leading to computers from 10+ years still being installed and used), their accessories or their intended usage function (i.e. just word? or AI learning and marine ecosystems?).
I would really love if the days of "over-provisioning" were over, but at least at my workplace, it is still a ways away.
@John Hanley, are you able to recommend a power usage meter? I may be able to ask my boss to see if we could get one. And is it a kind that is permanently installed or temporarily 'plugged' in somewhere (I'm assuming it's between the computer and the outlet). Thanks for all the help again. Highly appreciated from someone relatively new to this field.
I would really love if the days of "over-provisioning" were over, but at least at my workplace, it is still a ways away.
@John Hanley, are you able to recommend a power usage meter? I may be able to ask my boss to see if we could get one. And is it a kind that is permanently installed or temporarily 'plugged' in somewhere (I'm assuming it's between the computer and the outlet). Thanks for all the help again. Highly appreciated from someone relatively new to this field.
EnergyProfessional
Mechanical
An HVAC system has a life of what, 30 years? You don't design for a specific computer or equipment they happen to have today. You use a generally accepted and conservative design number. You'll need a good idea if the human/computer density is expected to be higher than ASHRAE or code assumes. If they have higher people density than ASHRAE/code assumes, you also need to up the ventilation. But you don't look up what type of monitor or computer they exactly use today or will use in 20 years.
We do the same with lighting and humans. You don't measure the metabolic rate of each person who currently works there, you use design numbers (ASHRAE has them for different activities). There will be past and future humans who will be outside the "standard" metabolic rates. That is a whole other discussion, but there is a trend of humans (in the US at least) to get bigger. So the metabolic rates likely will increase. who knows... use what ASHRAE or code requires and use some judgment to slightly increase those values if needed. If you get a good idea on maximum future occupant loads (and each occupant added adds computers, but doesn't add lighting), you'll be fine. As alluded, there is a maximum number of people that can be squeezed in a space. So there is your maximum.
Lighting still would stay at W/ft². but you could tie plug load to number of people instead of W/ft². More people use the coffee machine more, have more computers, print more etc. That way if you adjust for more people you will automatically adjust the plugload, ventilation, and people load.
if you micro-analyze every actual current setup, you will find 3 different office of the same size will have different loads today. Do you design each office differently? No... because next year they switch offices and the one person with the powerful workstation will move into the office that just had a tablet. Your design needs to be robust enough for "office use".
We do the same with lighting and humans. You don't measure the metabolic rate of each person who currently works there, you use design numbers (ASHRAE has them for different activities). There will be past and future humans who will be outside the "standard" metabolic rates. That is a whole other discussion, but there is a trend of humans (in the US at least) to get bigger. So the metabolic rates likely will increase. who knows... use what ASHRAE or code requires and use some judgment to slightly increase those values if needed. If you get a good idea on maximum future occupant loads (and each occupant added adds computers, but doesn't add lighting), you'll be fine. As alluded, there is a maximum number of people that can be squeezed in a space. So there is your maximum.
Lighting still would stay at W/ft². but you could tie plug load to number of people instead of W/ft². More people use the coffee machine more, have more computers, print more etc. That way if you adjust for more people you will automatically adjust the plugload, ventilation, and people load.
if you micro-analyze every actual current setup, you will find 3 different office of the same size will have different loads today. Do you design each office differently? No... because next year they switch offices and the one person with the powerful workstation will move into the office that just had a tablet. Your design needs to be robust enough for "office use".
- Status
Similar threads
- Locked
- Question