[stewie griffin]

Ok I've got to vent here and ask a question to the Chem E's out there. How hard is it to accurately charge raw materials by weight? I can't tell you how many times I've been asked to help fix a new process that is having problem X or Y only to find that, after examining the batch records, raw materials aren't being properly charged according to their stoichiometric requirements (sometimes under charged, sometimes over charged). It really frustrates me. The engineers prefer to charge materials using some other measurement (like pH, conductivity, etc). I have no problem using those as a backup but believe charge weights should be accurate as well.

So if we've got ~2000 gal tank and raw materials are usually added anywhere from 900 - 1500 lbs, what type of accuracy should I be expecting to get? +/- 10lbs? 20 lbs?

Mankind has been using weights and measures for thousands of years now so I have trouble understanding why this is an issue. Full disclosure, I'm an inorganic chemist but trained as a synthetic organic chemist, so maybe I just don't understand the practical aspects of doing reactions at scale.

[discodermolide]

I think that with the modern balances you can expect better accuracy than that. I was in chemical development in the Pharma business. As far as I remember we had balances which could weigh to 2 decimal places on a multi-kilo scale.
A robust process should be able to cope with some under or over charging, these limits will have been defined somewhere.

[curiouscat]

The relevant question here is how much are you willing to pay?
Weighing out 1500 lbs to a 2 decimal place accuracy is definitely possible. I suspect @Disco was making products that weren't exactly cheap. It's easier to afford the nicer gadgets there.

Some of the products I worked on sell for $20/kg and not much scope to fit in expensive equipment into these batch process.

Also, sometimes volumes are easier to meter out than weights, on scale. Is that a possibility for you?

I was looking at some of the Bronkhorst Catalogs and for your sort of capacities I found a Corollois MFC linked to a dosing pump that can go upto 600 kg/hr. The accuracy advertised is 0.2% but I tend to treat manufacturer data with a pinch of salt. I'd expect you'd get about 1% in typical use which is still about 6 kg of an uncertainty.

[discodermolide]

My products were never cheap, Ah but the quality was tremendous.

[eazye1334]

When I worked in the toner industry, we had 5000 gallon reactors fed from holding tanks that actually used flowmeters to calculate the amount of raw material that passed into the reactor. We only ever used the load cells as a backup because load cells are notoriously difficult. Due to adding on the fly, it took a skilled operator to know when to appropriately throttle down the flow to hit the targets because of the pipe distance between the valve and the reactor. Even so, our tolerances were around 15 kg over a 2000+kg addition and we rarely had problems.

[curiouscat]

My products were never cheap, Ah but the quality was tremendous.

I knew it! How much was your most expensive product?

[discodermolide]

It was priceless!

[stewie griffin]

Most processes I've been involved with charge by volume using flowmeters. Most of the time, as disco says, the processes are established and robust enough to handle slight variations.

What got me going yesterday was a new process that was tried on a scale (somewhere between pilot scale and normals scale) a few times without giving the desired results. As I said the charge weights of one raw material were listed as being +20lb over the theoretical need. Now maybe there's a good reason for doing so (it's not my process), but while I tried to find the reason for the deliberate overcharge, I told the engineers my first thought was to use the correct amounts. My resulting confusion from that convo on why it would be difficult to get accurate weights is what prompted my original post.

This particular process in question has dry solids as the raw materials, so both are weighed out and charged. Knowing that it's possible to get an accurate weight then makes me wonder how difficult is it to get all those solids into the tank (just in general, not necessarily as a potential solution for this process)?

@eazye1334
"because load cells are notoriously difficult"
What difficulties should I expect with the load cells. I've seen them used but never used them myself. Always just assumed they would be relatively screw-up free? 

[discodermolide]

Usually solids are just added through the "manhole". So it should not be hard to get it all in down to the last grain.
If they were using a solid addition apparatus then it becomes more difficult. They are usually driven with a screw drive with the solid in a hopper. If the stuff is hydroscopic them problems begin. Even if it's not the screw dosing can be problematic, especially if the stuff is lumpy.
Perhaps you should enquire what they used to add the solids.

[curiouscat]

This particular process in question has dry solids as the raw materials, so both are weighed out and charged. Knowing that it's possible to get an accurate weight then makes me wonder how difficult is it to get all those solids into the tank (just in general, not necessarily as a potential solution for this process)?

So long as they are free flowing it shouldn't be a big issue. How are you feeding the solids? A Hopper?

OTOH if they cake or stick it can be a nightmare.

[curiouscat]

What got me going yesterday was a new process that was tried on a scale (somewhere between pilot scale and normals scale) a few times without giving the desired results. As I said the charge weights of one raw material were listed as being +20lb over the theoretical need. Now maybe there's a good reason for doing so (it's not my process), but while I tried to find the reason for the deliberate overcharge, I told the engineers my first thought was to use the correct amounts. My resulting confusion from that convo on why it would be difficult to get accurate weights is what prompted my original post.

Its interesting though that you know you are 20 lb off spec. So that's not like a measurement error but a controller or dosing error.

Maybe the dosing lines are too long? There is some other lag in the feed system? The valve is too far from the tank?
Too much dripping after they close the valve? A leaky valve? Can't actuate it fast enough?

This could very well be bad design.

[eazye1334]

@eazye1334
"because load cells are notoriously difficult"
What difficulties should I expect with the load cells. I've seen them used but never used them myself. Always just assumed they would be relatively screw-up free?

In every case I've ever used them, they aren't. If the load isn't balanced properly, they will often be off in their readings. Strain gauges have a spring-like effect that leads to the load readout never totally settling, especially when adding raw materials. I also find they have a tendency to "stick" do to the physical (un)cleanliness of an industrial setting, leading to readings that are very off. And as dumb as it sounds, people forget to zero them out all of the time.

Granted, I've never used really expensive ones or used them in a non-industrial setting, so maybe those situations will see different results. But everywhere I've been, the load cells are always considered a last resort when all other measurement techniques have failed for one reason or another. They're nice to have, but I haven't found them to be terribly reliable.

[stewie griffin]

Its interesting though that you know you are 20 lb off spec. So that's not like a measurement error but a controller or dosing error.

Perhaps I should specify. The actual batch records called for an extra 20lbs of one raw material above its shoichiometric charge, and it was charged accordingly (or should I say, that weight was recorded as being charged by the operator). I finally tracked down the process chemist that developed this process and was told his reasoning for the deliberate overcharge in the process. It wasn't the most satisfying answer but at least now I know why it was done.

The point remained the same though, in that I ended up engaging the process engineers on how accurately we can charge dry solids by weight.
After further discussion today I found out that apparently they prefer to let the operators dump in bags of known weight. For example, if a process calls for 525lbs of material X, then the operators will add 10 bags (at 50lbs a piece) and then half a bag. How common is this practice?

As an aside, I appreciate everyone's feedback. The transition from grad school to process chemistry has been quite fun and surprising. So here's a general question put to you all:
If you're an engineer, what would you most want a new process chemist to understand, or what misconceptions does a new process chemist need to get cleared up? Or alternatively, if you're a chemist, what do you wish you knew prior to starting a career as a process chemist to make the transformation easier?

[discodermolide]

Last question first: As a development chemist doing lab. work to scale-up his own procedure in the pilot plant, after 15 years of this I would say always expect the unexpected. Your example says it all. Do the bags contain exactly 50lbs, no they don't, how do the operators measure half a bag? They can't. So I hope that process is robust. It is certainly not conforming to cGMP. Furthermore lofting 50 lb bags around cannot be good for their backs so they will eventually be off work with "back pain". Our balances had printers so the weights were actually recorded as part of the batch record.
I've seen valves being left open and solvent being pumped in at the top, you get a phone call "Dr Disco the reactor is not filling up, what do we do?".
Valves being fitted on the wrong way round on cylinders, i.e. the output valve was where the gas inlet valve was. This was discovered halfway through adding DIBAL-H. So we had to stop, call a mechanic get him to make a new valve fit it and somehow carry on, all the time the reaction is stirring away in the background.
Just a couple of examples.
To get your operators to alter their methods of addition or their practices in general you are going to have to change the philosophy of the plant which will not be easy and will meet with great resistance. But the operators, although usually a great bunch of people, will always find a way round this to make life easier for them.
Basically it means that you are going to have to stand there and watch over them and jump when you see something wrong, because if they can take a shortcut they will do it.
In Pharma our operators were trained in cGMP so I had the feeling they would not take a dump without a written procedure and then documenting it. However they still took shortcuts when they could, unless you were there at every critical phase of the process. Not with safety though, there they were good, but with things like running the temperature up too fast, or cooling too slow or fast. This got better when the equipment was computerised, but they still made "mistakes".
Get to know the operators, chat to them. Get the plant manager to give you a tour of the plant, ask intelligent questions, Show an interest in their problems, just generally be human not a boss. I found that this worked quite well, especially when I compared my approach to that of some of my colleagues, who never showed up to see what was going on, who, when they did acted like little Hitlers stomping their feet and giving orders.
Anyway I've ranted on enough, just always expect the unexpected.

[curiouscat]

The point remained the same though, in that I ended up engaging the process engineers on how accurately we can charge dry solids by weight.
After further discussion today I found out that apparently they prefer to let the operators dump in bags of known weight. For example, if a process calls for 525lbs of material X, then the operators will add 10 bags (at 50lbs a piece) and then half a bag. How common is this practice?

Here's how I think about this: If the process is relatively insensitive to X and X is not very expensive I see nothing wrong in "dump half a bag". You might be surprised how accurately a veteran operator can do this.

 OTOH, if the material is expensive or quantity critical to the process then they must weigh it. The point is operators will most often take a shortcut when they see themselves as taking irrelevant steps. I've seen managers that insist "every step is critical" but I think that's an impossible battle.

If you do have a 1500 lb batch the question to ask is how accurately must I specify a charge.

Operator time is expensive too. So sometimes it is cheaper to waste a lb or two of material than force an operator to jump through hoops.

I suspect the Pharma / GMP guys will give me a lashing for my blasphemy.