Nascent advanced nuclear technology industry should use only S.I. units
On the recommendation of a respected commenter, I have segregated a topic that was originally introduced as an afterthought at the end of another post.
Advanced nuclear technology developers, since we are starting an industry almost from scratch, should choose international (S.I.) units of measure at project development initiation. Some people who are interested in the technology but not active in the industry find it almost unbelievable that anyone has to make this recommendation. They assume that the action was taken long ago.
My experience tells me otherwise.
Ask an experience Areva or Westinghouse engineer or project manager to help you understand the enormous costs of converting nuclear plant designs from U.S. to S.I. units or vice versa.
Reject any advice received from Bechtel or other construction contractors to use U.S. units in order to ensure that their antiquated computer systems remain functional.
My guess is that the real aim of that kind of advice is to make work and require change orders in order to be ready for both domestic and international sales. Nuclear technology should be developed for the largest possible world market and to minimize the chances and potential consequences of silly “unit conversion” errors.
That’s enough from me. The floor is now open for commentary and questions. I’ll start by moving a few comments from the post from which this discussion was removed.
Reactor physics calculations have always used SI. The rest have used English units. Supposedly, this sprung from the fact that the physicists doing the reactor physics, being physicists already used SI while the engineers designing the rest of the plant always used English units.
I prefer SI, if only to avoid thinking about pounds mass vs. pounds force vs. poundals vs. slugs.
Nuclear already deals with quantities spanning magnitudes between 10E-24 and 10E23 without these idiosyncrasies. Plus international radiation units are based on SI.
Sounds like the industry would be in much better shape now had those reactor physics guys been willing to run the show back then instead of the engineers designing the rest of the plant. Hindsight is 20/20 though.
No they haven’t. The nuclear physicists made up their own units.
The barn is not an SI unit.
Not really. I have yet to see a course on electrodynamics beyond the most basic introductory level for undergraduates that uses the MKS-derived SI units. The preferred set of units for this subset of physics is the CGS-based Gaussian units. If anyone is curious, the reasons for this are explained in the “Appendix on Units and Dimensions” in Jackson’s Classical Electrodynamics, which I believe is still the canonical graduate-level textbook on the subject.
Totally agree. I *always* have to double-take and re-check how many rems are in a sievert (is it 10 or 100?), or millsieverts in a rem (is it 100 or 10?).
>The rest have used English units.
Here in England, we’re trying to give those things up, and have almost succeeded. Everyone fills up with litres rather than gallons. Miles, and miles-per-IMPERIALgallon survive in common use, but almost everyone nowadays has a feel for kilometres too and of course engineers and scientists have been trained in SI units for donkeys’ years.
What sort of unit is “donkeys’ years”?
I believe it’s approximately equal to twelve months of wet Sundays.
I was wrong when I said earlier that everyone uses the same units for time. Forgive me for having forgot about donkey years.
Whenever one of my fellow citizens say that metric it’s too complicated, I ask them how many inches in a mile and that shuts them up.
America has been exceptionally stubborn about changing to SI. Perhaps it’s so we can feel exceptional.
“I ask them how many inches in a mile …”
5280 × 12
The multiplication operation is left as an exercise for the reader.
Whoa, where did that 5280 come from? You have to leave the reader more to do.
Number of inches in a mile:
12 x 3 x 2 x 11 x 10 x 8
(accumulating, in succession, to feet, yards, fathoms, chains, furlongs and miles).
You never memorized that there are 5280 feet in a mile?
I don’t walk around knowing off hand how many inches are in a mile, but I’ve had that figure in my head ever since I was a kid.
There is a good reason why our teachers thought it was important to memorize certain facts. Sort of like using indexes in a computer program for fast retrieval.
I like 8760; though many nitpickers like to point out that it’s really 8766.
@Rod I have to use that 8760 number all the time. I will point out that time is consistently measured throughout the world. Can you imagine if someone used units other than hours or years?
As I’ve mentioned on this site recently, it’s really 8772±12.
The low end of the range just shows up a little more often.
The French experimented with decimal time in the years following the Revolution, but it didn’t take.
Am I the only one who needed maybe 5 seconds to come up with 63,360?
Yeah, but it took me 1 second to come up with 100,000 cm in a km.
I hope that you put those extra four seconds you saved to good use.
Reposting my original comment:
I am surprised that you even had to suggest using SI units as for any competent engineer their use should be practically axiomatic. Every time I read about NASA and other government agencies blowing millions of dollars because of a conversion error or I have to work with US unit bolts on my car I slap my forehead out of the sheer man-hour inefficiency of converting in a non-base 10 system. This is even ignoring the positive network externalities of a universal unit system that every else in the world uses. Using U.S. units on any project is literally a Homer J Simpson move:http://cdn.meme.li/instances/300×300/28951640.jpg
Please tell me no one in this industry has considered using US units. I am afraid though your suggestion suggests otherwise.
That Homer Simpson link should be http://cdn.meme.li/instances/28951640.jpg .
Seriously, how expensive / difficult can it be to hire a software engineer fresh out of college to write an API layer for that antiquidated computer system. Unit conversion is hardly the pinnacle of computer science. It’s quite deflating to think that parts of this industry once literally kings of the world, literarlly able to move mountains with the push of a button is now either unwilling or unable to complete this cheap and conceptually simple task.
However, then it occurs to me that there’s no better sign that the industry is ripe for disruption. That maybe outside of the anticompetitive dirty tricks of its competitors, the reason nuclear has not been able to gain market share over its competitors (particularly the capital cost challenge) is not due to any inherent flaws in the technology but rather flaws in the established players currently controlling it. This thought makes me actually project a very bright future for this industry in spite of all the forehead slapping moments such as the above.
Unit conversion for complex manufactured components, large systems, and extensive construction projects is a bit more complicated than you have described.
Think of all of the measurements that might be associated with a manufactured valve, including all of the close tolerances. Where does rounding begin? If you want to take advantage of some existing pumps or valves, how do you specify the spare parts associated with that equipment?
If you know enough about hand tools to understand why mechanics have both english and metric wrenches and then extend that vision to a power plant, you might begin to grasp the challenges.
That’s why now is the right time to make the switch for the nascent advanced nuclear energy industry. It has to be done before inertia sets in.
I agree with the general idea, but I have to ask (not being an engineer or businessman myself, so I’m out of my depth) – what about supply chains? Is it possible that some English units may need to be used because some part that has already been designed, tested, and approved would be far too expensive to re-design and re-test at a new, slightly different size because of changing unit systems?
For example, you not too long ago wrote about the challenge facing getting something that, IIRC, was called a “Canned Pump” approved for Westinghouse’s AP-1000 reactors? Now that that part is approved (I think? Or close to it), it seems like it would make sense for other nuclear designers to use the same part. What measurement standard is it constructed to? S.I. or English?
Are there other such parts that aren’t S.I.?
Seems like that’s the challenge, isn’t it? The supply chain?
A large, if not most, number of suppliers have both English unit and metric unit versions of their products for selling in the international market. For example, valve manufacturers such as Swagelok or Parker Hannifin will have near identical products for 1/4 inch pipe and 6 mm pipe.
On a different note, during the 1990’s I swore that the US kept both systems merely as a way to torture engineering students. We were always given tests with mixed units but no conversion chart(s) and we had simple calculators then. I always had to spend precious time converting units (we could pick which system) but I do think it flushed out some of the more careless students.
Rod, how about the “rod”?
Certainly, you must have a soft spot for that unit.
The rod or perch or pole is a surveyors tool and unit of length equal to 51⁄2 yards, 161⁄2 feet or 1⁄320 of a statute mile and one-fourth of a surveyor’s chain.
Interestingly, the “rod” is about the length of a PWR fuel rod.
You clearly have no idea about how Quality Assurance in the nuclear industry works. Let me put it to you this way, if nuclear power plants were designed and constructed by software “engineers,” there’d be a meltdown every other week, if not more often.
As you point out Bechtel’s computer systems as my API comments were directed at currently require US units for their inputs. Writing an API wrapper to enable the conversion of customer supplied inputs to US units really should not be a hard exercise, but instead of going through with it Bechtel rather impose even higher costs on their customers through that rather bad piece of advice to design in US units noted previously.
I don’t know about the advanced mini-LWR’s, but the Next Generation Nuclear Plant (NGNP) project has used SI units from the beginning — or at least AREVA’s High Temperature Gas Reactor (HTGR) project has. (AREVA is now the NGNP Industry Alliance’s preferred vendor for the reactor design.)
As I recall, this was a conscious decision that was made early on in the project.
Personally, I think the units thing is way overblown. Sure, it’s nice for a little consistency, but it’s a darn poor engineer who can only work in one set of units.
But when it comes to engineers, I’d prefer it if they’d spend less time worrying about systems of units and more time thinking about significant digits and how to write worth a darn — two areas where engineers are notoriously deficient.
The snafus that are commonly cited as a reason for everyone going metric (e.g., the NASA computer program that crashed the rocket/space probe/whatever because it used the wrong units) are not really caused by units. Units are merely the scapegoat. The real reason requires a little more thought:
What we have here is … failure to communicate.
That is, at the end of the day there was a combination of poor communication on one end and bad assumptions on the other end. This is not going to be cleared up merely by changing the coefficients (conversion factors) that are used to do the work. Sorry to disappoint the metric fans out there.
So while I completely agree with the advice that “nascent advanced nuclear technology industry should use only S.I. units,” I think that the arguments for it are rather weak.
US units are more common than most people imagine. US commercial nuclear units use US units. There is some dysfunction built in just for humor. For example, BWR Local Power Range Monitors read out in watts per square centimeter. That’s a pretty good expression for neutron flux because watts does relate to a certain number of fissions per second and each started with a neutron absorption. The Linear Heat Generation Rate (LHGR) for BWR fuel ( a GE number) is expressed in ( drum roll, please ) kilowatts per foot. Dang. And that bleeds over into Average Planar Linear Heat Generation Rate (APLHGR or apple hugger for the acronymically inclined) which is also in kilowatts per foot. Part of our dysfunction is our inbred ancestry. Many of us were BTU making Navy Nukes. We cherished our 3413 BTUs per kilowatt hour so much that we memorized it. Right next to 778 ft lbf per BTU. After we left the US Navy Nuclear family we hired on with traditionally fossil utilities to operate their shiny new reactors. (I’m old.) If you’ve been beer drinking with the gritty hand fossil folks, you’ve compared heat rate. Plants heat rate is in BTU per kilowatt hour. Coal energy in divided by electric energy out. The lower the better.
Anyway – feel free to cock all that up. I’ve had my 40 years of fun with Steam Tables and Mollier Diagrams in BTUs. Someone elses turn.
Engineer – hours are expensive and while any comptentent engineer can certainly work with any unit system, it takes a competent manager to understand why US units are just a terrible idea. Compare how long it takes to convert between SI units and the time it takes to convert between US units and then multiply by the number of conversions annually and I beleive you will find the increase in labor requirements staggering.
Are you schizophrenic?
First, you’re telling us that units are a problem that can be dissolved away by a clever, pimply-faced computer-science kid fresh out of college. Now you’re claiming that they’re an intractable problem that perpetually eats into productive time.
Well, which is it?
Sorry, but you can’t have it both ways.
Can you have a conversation beyond calling names?
It’s simple when humans are involved in the design process the cost of US units are prohibitive when computers are involved the cost is trivial. So the economic solution is for the computers to adapt to the humans NOT as Bechtel suggests to have humans adapt to the computers.
Certainly US units are more intuitive and I’ll certainly admit that I have no idea what temperatures in Celsius feel like thanks to my US education. However, when dealing with precision machinery, the last thing you want is people operating on intuition especially the labor. You want people reading numbers and comparing them to charts and as a side effect SI units also pushes IS labor out of its comfort zone onto following dials and charts to the letter.
Interesting what influence education can have. I’ve always thought that water freezing at 0 Celsius and boiling at 100 Celsius was fairly intuitive. But of course, change the pressure and the “intuitive” values no longer apply.
And in degrees Fahrenheit (and degrees Rankine) the separation between melting and boiling (at approximately 1 atm) is exactly 180 degrees.
Is 100 or 180 the better number?
While you think about this, let me ask … how many metric system proponents measure angles, compass directions, etc., in gradians (the “metric degree”) instead of degrees?
Personally, I’ve always found Fahrenheit convenient for weather, because (where I live) 0 is likely to be the coldest day of the year and 100 is likely to be the hottest (each of these is plus or minus, say, 5 degrees). So it ends up being a percentage scale between “coldest” and “hottest.”
Nevertheless, I find that having “0” mark the point when rain is likely to turn to sleet or snow and when water on the roads is likely to freeze is a useful feature of Celsius. The number “32” seems somewhat arbitrary.
100 is the better number, surely.
But measurements of angles in degrees, although non-metric, are at least universally recognised and almost as universally used. This is the key virtue when we’re talking about avoiding conversions to different systems.
(But what I *really* want is a universal switch to the FFF system (furlongs, firkins and fortnights). We should also measure volume in barn-megaparsecs (about 3ml).)
As far as the physical space is concerned and where does the rounding begin. Thanks to digital instrumentation decimals are cheap and there’s really no need to round off so soon so long as tolerances / error ranges are specified.
To drive my point home. When your children inevitably get into your tools compare how easy it is to get the metric sockets into their proper place compared to the fractional US sockets. I can certainly work out the order of fractions but the aggravation and the labor required is much higher.
When you multiply that additional labor requirements by the scale required for industry I’m certain you’ll find that US units place a significant additional cost.
The one downside I can find to the SI system is human estimation of measured quantities. However, I would say that thanks to the availability of cheap sensors now, the need for human estimation of quantities is almost non-existant.
“That is, at the end of the day there was a combination of poor communication on one end and bad assumptions on the other end. This is not going to be cleared up merely by changing the coefficients (conversion factors) that are used to do the work. Sorry to disappoint the metric fans out there.”
Certainly poor communication was a problem but again I come back to economics. The cost of communicating using two different languages is higher than communicating under a single language. In this case, had NASA decided to go straight SI this would have eliminated the need to convert in the first place and this not only reduces man-hour but also the opportunity for costly calculations errors just like this.
It’s a darn poor engineer who can only work in one language.
Personally, I think that it was a mistake when many engineering schools in the US dropped the foreign language requirement.
And it’s a darn poor manager that mistakes capability for necessity.
This link may prove enlightening…
Let’s take pipe sizes as an example. There’s ANSI (B36.10) and there’s DIN or DN (the metric standard). Look at this:
ANSI 2 inch sch 40 OD= 2 3/8 = 2.375 inch; ID = 2.067 inch
DN 50 std: ID=52.5 mm (= … Wait for it … 2.067 inch); OD=60.3 mm (= 2.374 inch).
so, there’s nothing inferior in the ANSI std, it’s just units
Computer programs: we use a lot of them where I work, and many are reviewed and approved by the US NRC. These were originally written in (gasp) FORTRAN. Changes have to be very carefully limited to avoid requiring NRC re-review; changing all the guts to take out the 32.2’s and divide by 144’s would certainly require a lot of work — these programs run to millions of lines of code. Have we written pre- and post-processors to input / output in SI? Of course we have. But does that make it a “metric” code? Of course not.
Units are just units, use whatever is convenient. The physicists use units where the speed of light is 1.000.
Oh, and you guys with two sets of wrenches are lucky. I have some old British machinery with Whitworth fasteners. That’s a third set of wrenches. And taps & dies. And thread gages. And I can’t buy nuts & bolts at Ace hardware, I have to send away for them, or make ’em on my lathe, which luckily has an inch leadscrew. Ironically, it was Mr Whitworth who championed the notion of fastener standards in the first place.
The Whitworth thread was designed for cast and wrought iron so it’s got a 55 degree angle, not 60 degree as all other modern threads do. Saying that a lot of Whitworth fasteners have US equivalents — the 1/4″ WW thread is the same pitch and diameter as 1/4-20 UNC and will work interchangeably.
And yes I too have old equipment that has Whitworth threaded fasteners.
I agree with you the guts of the code cares very little about US units or SI units, the impact is all on the user interface layer and I would argue the choice of system there matters very much to a company’s bottom line. The good news is that writing pre/post processing layers are not difficulty but apparently it’s too much for US construction firms to be willing or able to do so. They shouldn’t be surprised when they easily lose businesses to a competitor that does take this simple step.
Cavan, I don’t think you understood. The guts of the code cares intensely about the units because all of the constants in the equations are appropriate for the units of the variables. The property routines (temperature & density from pressure & enthalpy, etc.) are all written in the same system of units. Changing all that would be a horrible mess and mistakes would be likely, and it would take a lot of effort to find all the mistakes. That’s why, if you must show inputs and results in SI, it is easier to convert the SI inputs to “imperial” units, do the calculation in “imperial” units (with the code that has been used for 40 years) and then convert the output to SI. So, if you’re looking at the intermediate results, I mean the values of the variables while the code is running, they will be in feet, pounds, gallons, Btus and Fahrenheits. That’s why the pre / post-processors are a kludge, they force the analysts running the calculation to think in BOTH systems instead of just ONE system. Anyway, that’s my 2 cents.
I think we may have a double layer of misunderstanding. Please forgive me for not being specific in my comment. The cost to the computer of running variables inside the guts of the code in US units equals the cost of running variables in SI. The cost to the user to handle input/output in US units is huge compared to the cost of SI Units. Meanwhile the cost to change the guts of the code is huge but the cost to add a preprocessing layer is trivial. The solution is thus to write a preprocessing layer. I thinks this is not a kludge but actually good software design given the circumstances. Theres notthing wrong with code containing multiple layers of scope. Ideally though if we were starting from scratch wed want the guts of the code to be in SI.
I should note two that the preprocessing layer is modular. Got a customer insisting on some weird unit system I/O. Unplug our SI layer, plug in our weird system layer.
I think it’s actually quite easy to program a change units or, better still, give a user a choice of units. I’ve done that kind of thing before. An NRC review might cost a lot more. Or maybe finding a FORTRAN programmer is the issue? Most likely it’s just sheer inertia.
Doing the unit conversion math after the design has already been completed is not the only costly/time consuming part.
It’s also quite complex and expensive to change all of the tooling, specifications, spare parts, stocked materials like rebar and piping, etc. The implications on the supply chain can be immense.
Hence my prescription. Start in SI units and use them consistently. The NRC does not require Imperial Units; U.S. designers and project leaders like those who began the AP600/AP1000 design work made the choice. The EPR was being converted from SI to Imperial because it appeared to all concerned that the customers wouldn’t buy a plant designed, manufactured and constructed using S.I., but I’m pretty sure they would have made their acceptance more clear if the EPR sales/marketing team had explained the cost advantages of adhering to the internationally accepted units of measure.
I have always done most of my calculations in SI, starting in college. My least favorite English units usually have to do with heat transfer. British Thermal Units? Really?
Simon Aegerter translated my book, THORIUM: energy cheaper than coal, into German. But he said the hard part was translating the illustrative numbers, units, and calculations into units a European would understand. The book uses words like BTU, quad, acre, mile, ton, gallon. Also note that rad, rem, and Sievert are not SI units; one SI Gray = one watt-second per kilogram. [Sievert = Gray times a factor determined by a committee whose membership changes.]
I never understood why Americans use English units. Isn’t that like a battered wife who, after finally throwing out her violent husband, keeps his pictures on the wall?
If Westinghouse had originally submitted their AP1000 application for a British GDA in S.I. (instead of silly Imperial) we might be building AP1000s at Hinkley C instead of EPRs. Hinkley C has been a boon for anti-nuclear power activists and writers. A self-fulfilling prophecy: “nuclear power is too expensive”. I’ve lost count of the number of blogs and newspaper articles read over the last 2 years quoting Hinkley’s expense.
Why haven’t Americans ever tried to could the cost of lost exports? (or have you?) Maybe this silly Imperial unit thing a industry protection measure?
This particular American hasn’t counted the costs. I’m kind of lazy that way. Once I recognize that something is really freaking expensive and that there is a relatively straightforward way to avoid that something, I begin advocating the solution. I realize that many will expect an acceptable “proof” like Ms. Hanlon showed me how to do in high school geometry.
Even then, the requirement to take the time to write down each step frustrated me when the end result was exactly the same answer.
Having recently been a guest to the USA it was a walk down memory lane to re-engage with the Imperial system (as we call it in Australia). Personally I think the furlongs per fortnight is a great unit for velocity. Mid way through my undergrad engineering degree we changed from slugs, pounds mass and pounds force and went over to SI – it took us no time at all and freed us to engage on the concepts rather than the units.
Recent Comments from our Readers
The Clinton Nuclear Plant also in Illinois was shutdown essentially for almost 2 years before it was taken over by…
Good Podcast – Very informative One thing that was not discussed is how to deal with a particular fear that…
Renewables people are masters in marketing. Unreliable intermittent generators whose output is all over the place, and usually badly correlated…
Looking at their lineup, Westinghouse seems bound and determined to keep Gen IV in its “place” which is apparently the…
So they are developing a scaled down version of the AP1000, which is a scaled up version of the AP600,…