Hacker News Reader

Comment by rexreed | Hacker News Reader

How do you get so far into a project before you even realize that you have such a show-stopper problem? Move fast / fail fast works with software because iteration times are quick and penalty per iteration is low. But just building stuff and hoping it works and then realizing major problems doesn't work so well with real-world stuff. Couldn't they have "workshopped" this with non-functional item to see what the real world roadblocks would have been before a deep investment?

This is especially the case in high-risk unproven areas such as cleantech where the solution space is vast and potential problem areas significant. Any problem area can kill a project, so why are we progressing so fast to build first?

View on HN

Somehow you never quite get the real raw material when you ask for samples, no matter how careful you are.

I worked on a system once upon a time that fed steel balls into a ball mill. We tested it thoroughly in the workshop with the same steel balls used on the customer's site, but when we went to commission it, it jammed non-stop.

Turns out a hundred randomly selected balls won't contain most of the outliers that you'd get in even one tonne of balls, and when you're feeding 5 tonnes per hour, that's a lot of jams. We had balls with big craters in them, half-balls, balls with two halves offset by 50%, and everything in between. Not the kind of thing you could rely on rolling nicely.

Also, of course, when you ask a mill ball manufacturer for a sample, they might be inclined to send you the very nicest examples they can find, because they think you might buy their product...

Anyway, unless you're super careful about sourcing legitimate feed samples it's easy to think you're testing against the real product when you really aren't.

I found your comment fascinating (and frustrating). Dealing with bad inputs on a web form can be tricky. Dealing with bad inputs made of steel seems like a lot more work =].

What comes out of a steel ball mill? And why is steel made into (misshapen) balls before this process?

Oops, I meant to come back to this but stuff happened. It certainly was a lot of work to get it halfway reliable. I think they were working on adding a vibrating motor to keep things moving even if they weren't rolling - while I was there they just had a guy poke it with a bit of rebar every time things hitched up.

A ball mill is used to grind things into paste, usually for further processing. In this specific case it was a SAG mill (https://www.mogroup.com/portfolio/sag-mills/) and it was frikkin' huge, powered by a 10MW electric motor. It ground up ore-containing rocks from a mine which were then processed to produce lead and zinc. The balls roll around in the mill and help to grind up the ore, which I guess works better if they're round, which they will be before long due to wear and tear no matter what shape they go in. They come out as tiny ball bearings. It was quite an interesting project!

I mean, you're wrong... it clearly worked out for them so far and it only took seven weeks to fix this issue. It's taken longer to ship completed commits at Google than that.

Did it though? As far as I can tell they only have part of the problem worked out, and the rest of the problem may indeed be a show-stopper. How many more show-stoppers before they run into the same issues that bankrupted the $300M company? If you don't know what the show-stoppers are, then what is the best approach for plowing ahead anyways? Maybe sticking with wheat instead of corn?

They discuss a bit in the article, but the material they had prototyped with was very different than the material they tried to scale up with.

It sounds like their scaled up plant accepted a wide variety of materials, but this one specific input stream slowed them down.

So they weren't shut down while they figured this out, they just processed the material they knew they could process until they had a solution here.

Why would you prototype with a material that's substantially different from the real-world item? I read the article and this bit puzzled me. The differences clearly matter, and the $300M bankruptcy they reference in the article even makes that point.

Probably they want to handle wheat waste and corn waste long term. Seems like there should be plenty of both to go around

They want to be able to process both the material they prototyped with and the material they started with (straw) and the material they tried to scale up with (corn cobs), and probably any kind of plant matter that has less structural integrity than say, timber.

Initial testing was in San Francisco, where corn stover is relatively tricky to come by because corn is not widely grown in California, and agricultural residues cannot be brought into the state because of the bugs. Wheat straw was much easier to procure in-state.

Since encountering the differences, we tracked down the rarer corn growers in California and now use corn stover for testing as well.

If corn stover is a critical importance to your business and business model, wouldn't it be more effective, efficient, and beneficial to your business to be where the critical resource and information and experience is? Basically why not move to Kansas instead of stay put in California?

Come on. No one wants to live in Kansas. And it's extremely difficult to build in a rural setting. You try prototyping an experimental cutting-edge technology in the middle of a corn field 2 hours from the nearest Home Depot or metal fabricator.

John Deere might find that interesting. Now headquartered in Moline, IL (Iowa border), they started up and built their company exactly in the rural America (Grand Detour, IL), when Home Depot didn't even exist. Yes it was over 150 years ago, but there's nothing that prevents you from starting an agtech company where the resources are if they are in the midwest. The Kauffman Foundation, a major entrepreneurial resource is based in Kansas City, MO.

The wikipedia page for Wichita, Kansas claims Boeing, Airbus, and Learjet among others operate design and manufacturing facilities there. You should be able to find a metal fabricator and a home depot.

There are less drastic options than Kansas that would work fine. Chicago has an international airport. Indianapolis as well (though it’s not listed in Wikipedia, which makes me wonder how complete that is). St Louis and Minneapolis are major hubs. Large airports besides those I’m less familiar with, but if you’re planning to process corn waste what you need is rail access, not air. So then you can add a lot more of the midwest and still maybe be able to find Bulgogi for lunch. The Quad Cities, Peoria, Milwaukee, Des Moines, Omaha.

https://en.wikipedia.org/wiki/Indianapolis_International_Air... ?

That’s not a list of international airports, though. I’m unclear what page I was looking at now, but it only had about a dozen on it and IND was not one of them.

Disconnected data is just information, not knowledge.

https://en.m.wikipedia.org/wiki/List_of_international_airpor... and https://en.wikipedia.org/wiki/List_of_airports_in_the_United... both list KIND

>> You try prototyping an experimental cutting-edge technology in the middle of a corn field 2 hours from the nearest Home Depot or metal fabricator.

To be honest, to me it sounds like a dream job.

Oh for heaven’s sake. You can find a whole bunch of talented mech e’s and ag e’s all over the corn belt that already live there and like it. John Deere in Waterloo is just one name of many. My nephew works at Ag Leader (he’s an EE). There is a lot of ag engineering talent in the midwest, it is silly to try to recruit it here in Sili Valley.

This is like the joke about the drunk looking for his car keys under the street light because the light is better than whete he dropped them.

Because Charm is being run as a VC tech startup rather than a traditional industrial company and there's a recent movement of trying to do ag-tech in the bay rather than places like the Midwest or even central valley. Probably doesn't hurt that the CEO/co-founder already lives in the bay.

Why you wouldn’t do ag tech in Chicago is beyond me.

I've worked in agriculture my entire life, and seven weeks is a pretty good turnaround for fixing these types of problems. the next batch of issues they're likely to encounter is handling product with variations in moisture content that further bungs the system. Then it will be incorporating new feedstock from other crops.

Agriculture is the intersection of industrial mechanization and biological systems. Unlike traditional manufacturing, flexibility and efficiency is learned over time as situations are encountered that exceed previously theorized boundaries/ranges.

I have a grower who used gigantic wood burners for heat instead of natural gas. When I walked through his boiler room I noticed a wheelbarrow full of nails and other fasteners. He said 90% of his labour and headaches with that system were dealing with steel chunks in the feed stream, something they barely accounted for beyond adding a magnet when they built the system. Not everything can be planned for in advance.

“Did he just tell me to go fuck myself?”

“I believe he did, Bob.”

If you rely on material that is scarce in one location but widely available in another, wouldn't it make sense to test where the material is abundant..?

They did.

What do you think the article was about?

Ordinarily, I do not reply to these kind of comments, but this article is receiving a lot of them along these lines. The comment breaks down to "Why did you not predict everything that could possibly go wrong before you started this new tech project?"

I hope that it is clear that it would be impossible to identify every problem before field testing. Even in commercial manufacturing, you will run up against novel problems and have to engineer a solution onsite, even though we have had factories for about 250 years.

When they start testing in a different state or country, they will discover a new crop that breaks their system and will go though this process once again. But now, they are more experienced, and it will probably be easier.

Yeah it's definitely a bit peculiar. You'd think they could have done some more research into the makeup of the material they'd be processing, get a good understanding of the needs of the machinery...

Wheat straw is a common agricultural waste product across the US, and if they can solve processing that they still win. It's not that they tested A for a product that will only process B. They tested A for a product that will eventually process A and B and C and...

Scaling up means testing with new inputs at scale. So now they've done that?

I’m starting to believe the idea that long term zero percent rates have made MBAs in charge of almost everything completely careless to the historical realities of business where you have a product and process before investing tons into a system.

It seems to me that with “endless free money” we miss things like “this one input stream will be a nightmare, let’s not do it”.

Absolutely agree. Also, how is this firm based in Silicon Valley / San Fran where the customers are primarily in the midwest? If they were serious about this problem, shouldn't they be where their customers or resources are? Building something in isolation and then "shipping" it to where it will be used seems very inefficient and they're clearly missing out on valuable on-the-ground information and experience. This seems more like a university research project than a serious concern?

Wouldn’t the reverse question then be, why hasn’t someone in the Midwest done this already?

They may have already used that material for something else?

Like animal feed. Or tilling it into the soil to increase organic matter.

There are other companies working on similar things in the Midwest. It is a complex problem that farmers are very interested in because they see potential dollars.

https://www.bioenergy-news.com/news/frontline-bioenergys-iow... https://www.news.iastate.edu/news/2016/12/20/rapid

Our customers are technology companies based in the Bay Area, check out the home page: https://charmindustrial.com

We delivered 90% of permanent carbon removals globally last year: https://twitter.com/charmindustrial/status/14511691760646430...

Reply on news.ycombinator.com