Thread
@absmalhotra @JesseJenkins @TimMLatimer @Atomicrod @ramez @ETH_EPG @antunes_morgado
I have found this discussion on scaling and complexity and customization fascinating.
I think I can add some interesting perspective to it. A 🧵
I have found this discussion on scaling and complexity and customization fascinating.
I think I can add some interesting perspective to it. A 🧵
First some background. I'm a process engineer and work in the process design, retrofits and fabrication of major equipment for the chemical industry, in a specific technology area that is particularly complex and large scale. That is WAY up in the top right 'CoPS' box.
The way this paper describes things resonates with me, but all hope is not lost.
We cannot do the energy transition without a lot of stuff in the 'Type 3' areas. This includes Syn-fuels/chemicals, the materials that go into everything (Smelters!), Long duration Storage etc.
We cannot do the energy transition without a lot of stuff in the 'Type 3' areas. This includes Syn-fuels/chemicals, the materials that go into everything (Smelters!), Long duration Storage etc.
In my experience the easiest cost savings in the 'CoPS' space is by reducing complexity on the process end.
Eliminating entire systems or steps has a much larger impact than anything else.
Next biggest is consistent codes and standards.
Eliminating entire systems or steps has a much larger impact than anything else.
Next biggest is consistent codes and standards.
You cannot have any significant learning if the design concepts and details, if not the exact configuration, keep being changed under you. Especially while in progress.
That also impacts schedules and interest costs both through duration, and risk to the project.
That also impacts schedules and interest costs both through duration, and risk to the project.
Factory building equipment is still extremally valuable, even if we don't end out with conventional assembly lines. At a minimum impact, we estimate productivity is 3x higher in a factory than field assembling vessels.
This is due to things like:
-having permanent installed, unmanned cranes,
-being able to position the equipment in orientations that are optimal for fabrication, not for end use.
-No weather delays
-Large scale automated welding machines (Autogeneous, no bevel welding!)
-having permanent installed, unmanned cranes,
-being able to position the equipment in orientations that are optimal for fabrication, not for end use.
-No weather delays
-Large scale automated welding machines (Autogeneous, no bevel welding!)
-Optimal setups for QC. This is especially important in nuclear, not that chemical industry is light on QC...
In addition there are real SOAK benefits, even if there is not huge NOAK/SOAK learnings. Assembly sequences, build for fabrication and construction etc.
In addition there are real SOAK benefits, even if there is not huge NOAK/SOAK learnings. Assembly sequences, build for fabrication and construction etc.
This all leads me to some thoughts on how we can use this.
For the high complexity parts of the transition we can't assume learning curves will have a huge impact, so we can't rely on that in our designs.
We need to design for competitive SOAK price points.
For the high complexity parts of the transition we can't assume learning curves will have a huge impact, so we can't rely on that in our designs.
We need to design for competitive SOAK price points.
We can do this by
-targeting to minimize the plant complexity by reducing the number of systems and interactions
-Separate the site specific parts as more modular (like cooling)
-Keep components to sizes that can be factory build, and only site attached
-targeting to minimize the plant complexity by reducing the number of systems and interactions
-Separate the site specific parts as more modular (like cooling)
-Keep components to sizes that can be factory build, and only site attached
-Minimize site construction complexity, especially field welding.
-minimize civil works. This is CRITICAL. Complex civil is where things go sideways and never recover.
-Complete designs before construction. We need buy in to not go back and force changes on started projects!
-minimize civil works. This is CRITICAL. Complex civil is where things go sideways and never recover.
-Complete designs before construction. We need buy in to not go back and force changes on started projects!
The chemical (and FF power) industries have a number of internationally recognized standards (like ASME) that mean we don't have to start designs from scratch each time. We can use substitute components that meet the specs and the project specific performance requirements.
Nuclear desperately needs something more like this where the level of analysis can be constrained. At least for smaller plants that can't absorb the hours required.
In those cases, somewhat overdesigned substitutable components are much cheaper than exact performance match
In those cases, somewhat overdesigned substitutable components are much cheaper than exact performance match
but heavy engineering workload to confirm acceptability components.
-If you can throw steel at a problem to make it go away (especially if it saves a system or a processing step) do that. Steel is cheap!
-If you can throw steel at a problem to make it go away (especially if it saves a system or a processing step) do that. Steel is cheap!
On the chemical plant side, the issue is we are usually dealing with site integration issues, where there are different capacity, emissions, utility, product mix needs, and layout constraints. This leads to customization every time.
This will always be an issue with the Hydrogen economy! We need to be prepared to deal with it.
On that side, producing H2 in 'type' 1 or 2 production methods, that can be made cheap, and adding a buffer to keep the more expensive end users going steady will be critical.
On that side, producing H2 in 'type' 1 or 2 production methods, that can be made cheap, and adding a buffer to keep the more expensive end users going steady will be critical.
This is also going to be an issue to deal with for getting carbon - especially from bio sources.
It is in particular going to be an issue for getting the materials we need.
Smelters are the embodiment of 'CoPS', as are high performance polymers etc.
It is in particular going to be an issue for getting the materials we need.
Smelters are the embodiment of 'CoPS', as are high performance polymers etc.
Scale helps the cost effectiveness of these types of processes. On that side, we should not expect to have distributed industry. There will be hubs of industrial users, and we should design the H2 economy expecting that.
Anyway, if you made it this far, thanks for paying attention. I think this is a very important topic, and happy to discuss industry perspectives on or offline.
An addendum.
-try to keep novelty to one aspect at a time.
If you are using a new construction technique, use a familiar and proven design.
Don't change too many parts of the process at once, or MoC and process at the same time.
-try to keep novelty to one aspect at a time.
If you are using a new construction technique, use a familiar and proven design.
Don't change too many parts of the process at once, or MoC and process at the same time.
Also, economies scale for plant size is driven bottom up by equipment size. As each type of equipment reaches it's maximum size it need parallel sets.
There is no more scaling benefit from that aspect, and some anti-scaling from complexity - like flow balancing controls.
There is no more scaling benefit from that aspect, and some anti-scaling from complexity - like flow balancing controls.
So the scale limit is reached when too much equipment is impractical to be single unit, and you might as well split the plant into two independant trains rather than having parallel everything.
Mentions
See All
Ramez Naam @RamezNaam
·
Dec 13, 2022
This is a great thread, Jesse. Thanks.