• 0 Posts
  • 26 Comments
Joined 1 year ago
cake
Cake day: June 10th, 2023

help-circle

  • I’m a data engineer/architect and it’s the same over here, I get asked constantly “how can we stuff AI into this solution?”, never “should we consider using AI here? Is there a value?”, my view, people don’t understand their data and don’t want to put in the effort to understand their data and think that it’ll magically pull actionable insights from their dataswamp, nothing new, that’s been a constant for as long as I recall.

    Like I totally understand the draw of new and exciting, but there’s so much you can do with traditional analytics, and in my view you really need to have a good foundation before doing anything else.



  • I’m also thinking that way wrt to “we need more fast charging for EVs to work”, I recall that plugging into a standard outlet will get you something like 5-8 km an hour, slow charging is totally acceptable for most people’s usages. If you’re in an area where block heaters are the norm you already have outlets at parking spots, if I could commute to work and plug it in, covers most commutes in a 8 hour day, even those of us who rarely go in and live 70k away I’d be getting most of my range back. For the amount I drive, level 1 charging is more than sufficient.

    I think a compact with 2-300 k range would suit me just fine, would cover the odd longer trip and I’ll totally grab a rental for anything longer, like I already do it I need to move a fridge.


  • I have a large chunk of my colleagues who have little to no experience using CLI tools, and totally have found the last part to be true. In fairness, documentation is all over the place quality wise (I generally find Microsoft’s useful but I’ve totally had issues in the past with undocumented or vaguely documented features/dependencies). People will google their issues, and increasingly I’ve found it doesnt point you at the documentation directly, instead stack overflow or medium pages.

    I feel like there’s definitely some conceptual… Stuff for lack of a better word that’s an issue, I’ve seen a number of people focus on the execution instead of trying to understand what’s the issue and define it logically, when pressed they struggle to explain.


  • If you want a SBC, a lepotato works really well, supposed to be more performant than a 3B. I used as an alternate to a raspberry pi for a klipper setup, running armbian on it now.

    There are updated versions of it as well if you need more performance, but they’re cheaper than an equivalent pi and importantly, purchasable which was an issue when I was putting together that printer.


  • Crawler is possible, still need to get into the line though, I recall there being a few options for tethered camera crawlers meant for sewer inspection. Visual does have drawbacks, can’t really size defects, as far as I recall it’s difficult to get full coverage and cleanliness is even more important, and you’d general need the operations on that pipe to cease. Ideally you want your inspection regiment to allow you to know something’s coming and be able to plan for it, example if I start seeing vibration increasing on some bearings, I can monitor them and start planning for their replacement on a scheduled shutdown.

    No inspection is actually a totally valid mitigation plan for some assets. Criticality and failure consequences play a large role in that as well as the feasibility to inspect. Electrical devices for example follow random failure patterns and historically don’t really have a timeframe between failure initiation and functional failure that’s actionable, so a mitigation strategy I’ve seen done is something like hot spares if it’s critical. On the other hand, something that is inspectable but won’t result in high consequence of failure (death and injury are the things that are usually weighted heavily) it might not be worth inspecting either, it’s all about trying to get the most out of limited maintenance budgets.


  • I did smart pigging and challenging pipeline stuff years ago and reliability engineering up to a few years ago, also got out of o+g for similar reasons.

    Totally agree and just adding on to all that, even if it was steel, I’m super willing to bet it’d be impossible to run a normal pig through, so much infrastructure is just full of diameter changes, unbarred tees, really tight back to back bends etc, I can only imagine the challenges to pig a line like that, let alone costs involved with specialised tooling and support work, I know some people who did a short run through a downtown core on a gas main and that needed hot taps, road closures and a really special pig for what was less than a kilometre.

    Supposing it could be pigged without blowing up their entire maintenance budget, I wouldn’t want to touch any of the water coming out of that line during operations, so you’d also interrupt water service for a while, having a solid reaction plan really would be one of the best solutions.


  • A lot of industry does use grey water or untreated water for cooling as it’s substantially cheaper to filter it and add chemicals to it yourself. What’s even cheaper is to have a cooling tower and reuse your water, in the volumes it’s used at industrial scales it’s really expensive to just dump down the drain (which you also get charged for), when I worked as a maintenance engineer I recall saving something like 1m cad minimum a year by changing the fill level in our cooling tower as it would drop to a level where it’d trigger city water backups to top up the levels to avoid running dry, and that was a single processing line.





  • I did a student project for server room HVAC fans being annoying back in uni, targeted reduction in those annoying or peak frequencies was a totally acceptable outcome as to not disturb operators (was for a simulated patient in the attached hospital). I’m not an acoustic engineer, so obviously take what I’m saying with a grain of salt (did do a lot of safety and risk work though), making things less annoying is perfectly valid if they’re not already harmful to your hearing in the first place.

    What’s cool to me is that it’s just printable, so in theory super accessible and anyone could iterate on it if they desired (assuming it gets open sourced)