Wednesday, December 6, 2023

Secondary Costs of Production (Rabbit hole alert)

This post was requested by a reader in the comments section of another post.

Please forgive me if it becomes too arcane.

Industrial Engineering

The discipline of Industrial Engineering, at its core, is consumed with the calculation of "costs" of production and using strategies to minimize that cost.

A core-concept of Industrial Engineering is that WHERE a "part" is scrapped out makes a huge difference in the cost of that scrapped part.

Picture in your head our North Korean cookie factory. Suppose cookie-girl dropped the entire day's production of speculative cookies into a puddle of manure-tainted water at the doors of the factory where she intended to sell the cookies.

Now consider the same cookie factory but the cook opens the container of corn-flour for the day's production and determines that it has been hopelessly polluted with manure-tainted water.

In the first example the following resources were wasted:

  • The time and calories it took cookie-girl to walk to the factory
  • The time and calories it took for cookie-girl to cook the cookies
  • The "rent" or opportunity costs for the griddle.
  • The time invested in finding, harvesting and transporting the sticks to fuel the griddle.
  • The container of corn-flour used as the raw material to make the cookies.

Compare to the second example where the only "waste" was the corn-flour that would have been used to make the cookies.

Because of this, one of the major strategies traditional Industrial Engineers use to minimize waste is to SORT. If a particular lot of incoming material (or labor) is likely to result in scrap later in the process, then red-tag the material and send it back to the supplier*.

Part of the lingo is "inverted funnel" thinking. Sort most aggressively at the start of the process and then progressively less aggressively later on when product has more "value added". The justification is that variation in incoming material and at the subcomponent-level creates scrap later on so from a wholistic standpoint it is economical to be very aggressive early-on and coast later in the process.

This was the dominant way of thinking about Industrial Engineering in the US until the late 1970s for a couple of reasons. One was that consumers were not very picky. Any car or house or washing machine was better than no car or house or washing machine. The other factor was that organized labor was dominant and militant. It was considered good strategy to have multiple suppliers so when one went on strike the other suppliers could pick-up-the-slack and the factory could keep running and keep producing.

In the chaos of famine and social unrest, this will be the most appropriate model for managing "secondary costs' because any cookie is better than no cookie.

Edward Deming or "Japanese" Industrial Engineering

While Deming/Japanese Industrial Engineering LOOKS very different from traditional Industrial Engineering, it has similar roots.

Suppose it was conceptually possible to dissolve the walls between your factory and the factory of your supplier(s). In the case of our cookie factory, that would mean that they had "virtual control" of the corn-fields, the storage sheds and the people running that operation.

Then, rather than sorting the bags of incoming corn or corn-flour they could drive the sorting back even farther in the process.

For example, mold is a major source of contamination in corn because molds (fungi) produce nerve toxins that can make you vomit and they produce toxins that cause cancer and they produce toxins that impact taste.

Some strains of corn are more vulnerable to mold than others**. The highest level of sorting would be to red-tag the strains of corn most vulnerable to mold.

Cultural practices influence mold. Planting the seeds too closely together means the corn does not dry off as quickly after a rain. Setting standards for planting density is a no-brainer since corn that is planted farther apart responds by producing bigger ears of corn.

Harvest practices influence mold. Leaving it in the field too long results in exposure to weather/rain and it can mold in the field. Harvesting it too soon means it is not dry enough and it will mold in storage.

Storage practices influence mold. The storage building or crib's design. Position of the crib matters because trees can block sunlight and air-flow. Vines growing on the sides of the crib can inhibit air flow and so on.

The Japanese were able to pull this "dissolving of walls between factories" because they did not have 8 suppliers for each item. It is irrational in the multi-supplier scenario for the supplier to invest heavily in improvement because the initial costs means they will lose sales volume (perhaps even all of their sales) to GM or Ford or Maytag.

The Toyodas of Japan picked their one best supplier for each commodity and said "We will send you all of our business as long as you play the game our way. We will provide guidance in where you need to improve. We will loan you technical expertise. We will pay slightly higher prices at the beginning of the process because it is not in our best interests for you to go out-of-business."

"BUT...we will demand lower prices as the quality improvements come on-line because you will be manufacturing less product in total because you will be have less scrap to destroy or (or rework)."

Adding inventory is a response to waste. Inventory masks or enables waste.

Some industrial thinkers believe that the biggest gain of Just-In-Time inventory is that it revealed those processes, practices or variables that created the most scrap. 

For example, a factory might run high inventories because they had a historical practice of running a reamer until it broke. The reason it broke was because it was dull or chipped and the torque required to turn the tool increased. "...dull or chipped..." implies that it was making scrap for the last 10 or 100 or 1000 parts.

The broken tool created waste in the unscheduled downtime required to replace the broken tool-bit. The pulsing scrap-rate meant material should be quarantined and sorted. The lack of smoothness in production rate meant that a large inventory of finished (expensive) parts had to be held in inventory so the production at the customer's plant was never impacted.

Stripping inventory out of the system (JIT) meant that somebody had to pay attention to what was driving the scrap. Once it became apparent that most of the scrap was created by reamers that were about to break then it is pretty simple to record how long the reamers last (record number of parts-to-failure for the next 10 reamers) and then start scheduling reamer changes BEFORE they break...say at 90% of the life of the reamer that died soonest.

Or, if the factory's management is in love with electronic controls, monitor the current draw of the motor driving the reamer spindle (i.e. the torque the motor produces) and when the current draw tickles a certain level to trigger a reamer-change. Dull tools require more torque to cut the metal.

The advantage of the replace-to-a-schedule is that the Plant Manager can require that the dull tool bits that were used be kept in a drawer and sent out for resharpening every Monday. He can open that drawer on Wednesday or Thursday and he should be able to count the tool bits and verify that the tool-bits are being replaced as required at the proper rate. He can visit on Monday afternoon to verify that they are being sent out for sharpening.

The control's based approach has the weakness of a lazy operator gaming the system by squirting oil on the tool to reduce the torque or to hit the reset since the "tickle" is probably a "X-number in a row" algorithm so the tool doesn't shut-down every time it encounters a casting or forging with a hard-spot in it.


The traditional model of Industrial Engineering will be most applicable in the chaotic, kinetic environment of rapid social decline. Cookie-girl never knows who is going to show up at the door of her hut with a bag of corn.

The Deming or Japanese model functions very well in a high-trust environment but can crap-the-bed when its foundational requirements are not met.

Note to Unknown in comments: I am not sure if this is what you are looking for.

*The supplier then parks the bin of red-tagged material in a corner and ships it back on second-shift. Or, if the parts were bounced by the second-shift inspector, the supplier puts a layer of known-good parts on top of the bad parts and ships it on first shift knowing that the first-shift guy is lazy and always pulls parts to inspect from the top of the bin.

**During the drying-down part of the corn production cycle, the ears rotate on the stalk from tip-up to tip-down. The husks "shingle" to shed water in the tip-down orientation. The timing of when that rotation occurs influences how the corn dries down and its susceptibility to molds. That timing is controlled, in part, by the genetics.


  1. I worked in print manufacturing for 20+ years, the JIT craze swept through there, too.
    It had some advantages, but the biggest weakness I saw (as pertains to future sp0rkiness) was most implementations did not plan for hiccups or backups.
    Since we were in Rochester NY, you used Kodak printing plates and chemicals (Kodak HQ = Roch). We negotiated a hefty consignment allotment and started to roll. Business picked up, and Kodak had a problem at their plant in Colorado.
    Now we're juggling with a 4 week lead time to get plates from Europe. We only stocked 2 weeks worth of inventory for consignment. That was fun for a couple months... Then their coating line in Europe had problems! Coating line in Japan was 6 weeks boatride away! And this supply problem was now global (Kodak only had 5 sites worlwide they manufactured printing plates at, and 2 were shut down!)
    We were able to switch to a secondary manufacturer fortunately, but we had to change to a new chemical process and recalibrate EVERYTHING. That 'all your eggs in one basket lesson' stuck with me.
    I get the advantages to optimization, but don't like the fragility that type of operation seems to have. I like Talib's concept of anti-fragility soooo much better (though harder to implement).

    1. Different horses for different courses.

  2. This is a posting worth several readings. It explained coherently things I knew at a gut level but never articulated. You are good at that.---ken

  3. When I worked for a small defense contractor, we tried whenever possible to have at least two qualified sources for any component. There were a few components where we had to use a single source specified by the DoD. That redundancy saved us more than once from failing to meet the contract schedule. Granted the second source took some time to come online when we needed them to produce, but the fact that we had already qualified them via First Article Testing, etc, shortened the down time considerably. The higher cost for the components from the standby supplier was less than the cost of defaulting on scheduled deliveries to the government.

    The fact that we had already done an FAT provided insight of what areas of their process were most likely to produce a quality escape and allowed us to better monitor for defects.

    I'm a firm believer in having a "plan B" whenever possible. At my next employer, there were a lot of items where we did not have a secondary source. It caused us a lot of grief with customers wanting to know when they would be getting their remanufactured or repaired item back as they often had few or no spares on hand. A locomotive that is out of service waiting on parts is an expensive, heavy paperweight. The railroads could also charge back some of the repair cost if we did not meet the turnaround time in the contract. Unfortunately the bean counters did not want to go to the expense of qualifying a second source, especially for components made by another division of the parent company. That ended up costing them several customers when we couldn't meet production deadlines.

    1. The fundamental flaw of bean-counters is that they believe their business MODEL is reality rather than an abstract, flawed, stripped-down simulation of reality.

      To make the model manageable, they insert shortcuts and sleights-of-hand, they leave stuff out and they "account" for some costs by using more convenient proxies for that cost and so-on and so-forth.

      Then the bean-counters punish those of us who swim in reality when we cannot make the outcomes conform to their abstractions.

  4. Sourcing is a huge issue in my work area. Often times we have no choice but to use a single supplier which sets up its own set of risks and planning - and that was before The Plague, which completely put timelines on their head. Things that were 3 months became 6 to 9 months and somethings (medical tubing, for example) were simply not available.

    The JIT model assumes high trust but implicit in the construct is also free trade and reliable supply chains . If The Plague showed anything, it is that the second - reliable supply chains - can fail for the seemingly most innocuous of reasons. I can only imagine what an actual serious event would do.

  5. ERJ, I am that unknown commentor. Your essay touches on my request quite well. WRT to inventory, you have reminded me of some of those hidden costs.

    CoyoteKen @ 5:52 sums up my thoughts.

    I had been a small business owner most of my career. I was so very often concerned with overhead. Actually, my perspective was that to an inordinate level I fretted on the costs. I just wanted to make stuff, ya know.

    My perspective as a small business was different - often worlds apart, it seemed - than the mid to large industrialist. This seemed incongrous as we were all wrestling the same beast. Same game, different levels.

    ERJ, I thank you for your thoughts. Too, the readers you attract who offer their own expertise.

  6. Great post. Small manufacturer here, never had any biz training.
    Somehow made it work. I have one sub for finishing, the only one around- try opening up a painting shop these days.
    The previous post on layering of permitting process's was relevant , apparently the feds have some new form added for corps. It's like the entire country is having a bad dream, running through waist deep mud.


Readers who are willing to comment make this a better blog. Civil dialog is a valuable thing.