Latest / Mechanical Engineering Made Simple / (#20) Lean Principles -Drag the Factory Floor to the Engineering Office

Transcript

(#20) Lean Principles -Drag the Factory Floor to the Engineering Office · Mechanical Engineering Made Simple

• 0:00
  Welcome back, design junkies, to mechanical Engineering Made
• 0:03
  Simple. So today we're going to try
• 0:04
  something a little different. We're taking lean principles,
• 0:07
  you know, this stuff usually associated with the factory
• 0:09
  floor and dragging them kicking and screaming into the
• 0:12
  engineering office. We're diving into lean
• 0:15
  engineering. Basically, how do you apply that
• 0:18
  relentless focus on cutting waste and boosting value across
• 0:21
  the entire product development process?
• 0:23
  Exactly. It's about seeing the whole
• 0:25
  operation from that first glimmer of an idea all the way
• 0:28
  to shipping the product as well. A single complex system.
• 0:32
  Right, not just optimizing isolated parts.
• 0:35
  Because historically, yes, Lean was about streamlining the
• 0:38
  assembly line. But we've come to realize that
• 0:41
  the biggest cost commitments, the really impactful ones, they
• 0:44
  happen way before anyone picks up a wrench.
• 0:46
  So our core mission here is really about accelerating the
• 0:49
  whole cycle time, getting rid of waste, yes, but also setting up
• 0:53
  really efficient standard processes.
• 0:55
  And this is where you start hearing about things like
• 0:58
  integrated product and process development IPCD.
• 1:00
  IPPD sounds a bit like jargon, maybe.
• 1:04
  Let's make it real. We're talking about pulling
• 1:06
  everything together. Design specs, manufacturing
• 1:09
  needs, supply chain logistics right from the get go, aren't
• 1:13
  we? Pretty much integration from day
• 1:15
  one. OK, so let's start at the
• 1:17
  beginning. Product development or PD?
• 1:20
  What is it? Well, the way I see it, PD
• 1:22
  starts the moment you spot a gap in the market, some opportunity,
• 1:26
  and it doesn't really stop until that product is being made, sold
• 1:30
  and crucially, delivered successfully.
• 1:33
  It's the whole 9 yards. It is the whole marathon.
• 1:36
  And if PDO and marathon, the main thing you're actually
• 1:38
  creating along the way is, well, it's uncertainty reduction
• 1:41
  uncertainty. Reduction.
• 1:42
  OK, unpack that a bit. Think about it.
• 1:44
  Every activity in product development, every design
• 1:46
  choice, every simulation, every prototype test, it's
• 1:49
  fundamentally about gathering information to kill risk.
• 1:52
  Every dollar spent, every hour invested should be chipping away
• 1:55
  at what you don't know. So as time goes on, that risk
• 1:58
  level should be dropping sharply, while the value you're
• 2:00
  locking in should be climbing just as fast.
• 2:03
  That makes sense, but it raises the question of where this
• 2:05
  uncertainty actually comes from. Risk is about the unknowns.
• 2:08
  Right. And engineers, project managers,
• 2:11
  they're used to dealing with the known unknowns.
• 2:13
  You know, the tests they know they need to run, the data
• 2:15
  points they know they're missing.
• 2:17
  That's part of the plan. But the real curve balls, ah.
• 2:20
  Those are the unknown unknowns, the things you didn't even know
• 2:23
  you didn't know. Maybe an unexpected material
• 2:26
  interaction, a sudden shift in regulations, or a competitor
• 2:30
  coming out of left field. Lean PD tries to use really fast
• 2:34
  feedback and concurrent work to surface those nasty surprises as
• 2:39
  early as humanly possible. And timing is just, it's
• 2:43
  everything here, isn't it? Because design so
• 2:46
  interconnected, those early decisions have these huge ripple
• 2:49
  effects. I remember seeing this classic
• 2:50
  graph. Oh, that 3 curve graph.
• 2:52
  It's practically scripture in systems engineering.
• 2:54
  It shows ease of change plummeting almost immediately
• 2:57
  right there in the conceptual phase.
• 2:58
  So making changes is cheap early on.
• 3:00
  Dirt cheap you can pivot the whole design for almost nothing,
• 3:05
  but flipside cost incurred the money you're actually spending
• 3:08
  and more importantly, life cycle cost committed the LCC.
• 3:13
  They spike upwards incredibly steeply right then during
• 3:17
  preliminary design. An engineer choosing, say, a
• 3:20
  slightly cheaper component today might unknowingly commit the
• 3:23
  company to hundreds of thousands in extra maintenance or unique
• 3:26
  spare parts over the next decade.
• 3:28
  Wow. So waiting to resolve
• 3:30
  uncertainty doesn't just cost a bit more, it costs exponentially
• 3:33
  more later on. Exactly.
• 3:35
  The penalty multiplies, which means you absolutely need tools
• 3:38
  and methods that force collaborative, informed
• 3:40
  decisions early, locking in the right LCC profile.
• 3:43
  And that leads us naturally to things like integrated
• 3:46
  concurrent engineering, ICE and it's sort of evolution, real
• 3:50
  time concurrent engineering RTCE.
• 3:51
  Right. ICE, which got its start at
• 3:53
  places like JPL, isn't just about having more meetings.
• 3:56
  It's fundamentally about doing the actual design work together
• 3:59
  simultaneously, not just. Status updates.
• 4:01
  No, not at all. You've got stress analysis,
• 4:04
  thermal, electrical, manufacturing folks all working
• 4:07
  in parallel. The magic ingredient is usually
• 4:10
  a central server, a single source of truth.
• 4:12
  All the design data flows through it.
• 4:14
  So if you're the thermal engineer, you see the structural
• 4:17
  engineer's latest bracket update the instant they committed.
• 4:20
  No waiting for drawings, no outdated information.
• 4:23
  I can imagine that's maybe a bit unnerving for engineers used to
• 4:26
  working in silos, but the benefits must be significant,
• 4:29
  That instant visibility. It changes everything.
• 4:31
  First, it drastically cuts down information lag and the blame
• 4:35
  game that comes with it. 2nd, and this is crucial, it shifts
• 4:39
  decision making. It moves away from subjective
• 4:42
  cause and, you know, design by seniority, where the person with
• 4:46
  the fanciest title wins towards objective, data-driven choices.
• 4:50
  How so? Well, if the structural engineer
• 4:53
  uploads a design and the thermal simulation running automatically
• 4:56
  flags it as overheating within minutes, the data makes the
• 4:59
  decision. It's not about who argues best.
• 5:02
  And the speed improvements reported are, well, they're
• 5:04
  pretty staggering, some teams getting new designs done in
• 5:07
  hours instead of weeks. Yes, the metrics are often
• 5:10
  dramatic. We've seen lead times for
• 5:12
  standard design tasks cut by a third or more, and that ability
• 5:16
  to generate a viable new concept design in, say, 4 hours versus 4
• 5:20
  weeks. That's not just faster, it's
• 5:22
  strategically different. It means you could explore maybe
• 5:25
  4 completely different architectures in a single day.
• 5:28
  Instead of spending a month just validating 1.
• 5:31
  You cover the design space much more effectively.
• 5:34
  Finding the global optimum, not just the first acceptable
• 5:37
  solution. Precisely.
• 5:39
  And this drive for efficiency also brings up another key lean
• 5:42
  idea, variability reduction through design for reuse.
• 5:45
  OK design for reuse, like using standard parts or making things
• 5:49
  symmetrical. Exactly that kind of thing.
• 5:51
  Multi use parts. For example designing A
• 5:53
  component so the same part works on both the left hand and right
• 5:55
  hand side of an assembly. Maybe just by flipping it over
• 5:58
  or making it a true mirror image?
• 6:00
  Seems basic maybe? It seems basic, but the
• 6:02
  downstream effects are enormous. Fewer unique part numbers to
• 6:06
  manage, which means lower inventory costs, less tooling
• 6:10
  complexity, reduced chance of assembly errors.
• 6:13
  Think about the person on the assembly line.
• 6:15
  If the parts are symmetrical, they can't install it backwards.
• 6:17
  Fewer mistakes, less rework, lower cost.
• 6:21
  It cascades. Right.
• 6:22
  And to do that effectively, you can't just think about your own
• 6:25
  design task. You need to consider
• 6:27
  manufacturing, assembly, even maintenance down the line.
• 6:31
  This forces a wider view. It absolutely does.
• 6:33
  It forces you to zoom out and think about the entire system,
• 6:37
  the extended enterprise. OK, the extended enterprise.
• 6:40
  Let's define that. A lean enterprise isn't just the
• 6:42
  company itself. It's this integrated entity that
• 6:45
  creates value for all its stakeholders, this whole
• 6:47
  network. Stakeholders.
• 6:49
  That word gets used a lot. How are we defining it here in
• 6:51
  this systemic lean context? Good question here.
• 6:55
  A stakeholder is basically any group or individual who can
• 6:58
  affect or is affected by the enterprise's value creation
• 7:01
  processes, directly or indirectly.
• 7:03
  That sounds incredibly broad. Customers, employees, suppliers,
• 7:09
  sure. But regulators?
• 7:11
  The media? Local communities.
• 7:13
  Yes, potentially all of them. Which leads to the next critical
• 7:16
  question. How do you figure out who
• 7:19
  actually matters most? You can't cater to everyone
• 7:21
  equally all the time. So how do you prioritize?
• 7:24
  How do you map their relevance? Or what's the term?
• 7:27
  Salience. Stakeholder salience.
• 7:29
  It's generally seen as having three key attributes.
• 7:32
  First is power. Power meaning.
• 7:34
  Influence. Exactly, the ability to make the
• 7:37
  enterprise do something it wouldn't otherwise do.
• 7:39
  This could be coercive power like a regulator threatening
• 7:42
  fines, utilitarian power like a major customer threatening to
• 7:45
  walk, or even symbolic power like negative media coverage
• 7:48
  hurting your brand. OK, power, what's #2.
• 7:51
  Legitimacy. This is about perception.
• 7:53
  Is the stakeholders claim or action seen as desirable, proper
• 7:56
  or appropriate within the norms and values of the organization
• 7:59
  or society? So does the stakeholder have a
• 8:02
  right to make this demand? Sort of.
• 8:04
  A union demanding safer working conditions usually has high
• 8:07
  legitimacy. A competitor spreading false
• 8:09
  rumors has very low legitimacy. Got it.
• 8:12
  Power, legitimacy and the third attribute.
• 8:16
  Criticality. This one often involves time
• 8:19
  sensitivity or the sheer importance of the claim to the
• 8:22
  stakeholder. OK, explain criticality a bit
• 8:25
  more. Think of a small supplier who is
• 8:27
  about to go bankrupt if you don't pay their overdue invoice
• 8:30
  today. That's high criticality due to
• 8:32
  time sensitivity. Or think of a community group
• 8:35
  demanding you address toxic runoff that's vital to their
• 8:38
  local environment. The importance of the claim
• 8:41
  itself gives it criticality. So power, legitimacy,
• 8:44
  criticality, you use these three factors to sort of triage the
• 8:49
  stakeholder network and figure out who needs attention when.
• 8:51
  That's the idea, and it helps make sense of that complex web.
• 8:54
  And there are conceptual models to help visualize this.
• 8:57
  Like the one mentioned in the research the the water drop
• 8:59
  model. Yes, the water drop model.
• 9:01
  It's a qualitative tool, not quantitative, but it helps
• 9:04
  picture the relationships. Imagine the company as the point
• 9:07
  where a stone hits the water, the first ripple right next to
• 9:11
  the impact point. Those are your primary
• 9:13
  stakeholders, key suppliers, your core employees, your most
• 9:17
  loyal customers. They have direct, frequent
• 9:20
  interactions and high interdependence with the firm.
• 9:22
  And the ripples spread outwards. Exactly.
• 9:25
  Those primary stakeholders have their own stakeholders, their
• 9:28
  suppliers, their employees, their communities.
• 9:30
  These become your secondary stakeholders.
• 9:33
  Think government agencies, the media, activist groups.
• 9:37
  Their influence on your firm is often mediated through those
• 9:40
  primary stakeholders. A new regulation secondary
• 9:43
  impacts your supplier primary, which then impacts you.
• 9:47
  So it shows these nested recursive relationships.
• 9:51
  Your connection with the supplier isn't just about price,
• 9:53
  it's influenced by their relationship with their bank or
• 9:56
  their local government. Precisely, it highlights the
• 9:58
  interconnectedness of the whole system.
• 9:59
  Which actually provides a really smooth transition.
• 10:02
  If the extended enterprise is the sensitive interconnected
• 10:05
  system, then managing the physical flow of goods through
• 10:08
  it. The supply chain also needs a
• 10:10
  systemic approach to avoid creating chaos.
• 10:13
  Absolutely. Which brings us squarely to
• 10:15
  Hijunka. Hijunka, a Japanese term right
• 10:18
  from Toyota Production System. That's right.
• 10:20
  It literally means leveling. It's the process of smoothing
• 10:24
  out production, both in terms of overall volume and the mix of
• 10:27
  different products being made. It's fundamental to TPS
• 10:30
  stability. And it's designed to combat.
• 10:32
  Yeah. What's that classic supply chain
• 10:34
  problem? The bullwhip effect.
• 10:35
  The dreaded bullwhip effect, Yes hi Junka, is the antidote.
• 10:39
  The bullwhip often happens because of perfectly rational
• 10:42
  local optimization decisions that cause massive upstream
• 10:46
  volatility. Give us an example of that local
• 10:49
  optimization causing trouble. OK, imagine a trucking planner
• 10:52
  for a regional distribution network.
• 10:55
  Their main performance metric might be truck fullness to
• 10:58
  minimize transportation cost per unit.
• 11:01
  Seems sensible, right? Yeah, you want full trucks?
• 11:03
  So to get full trucks, the planner waits until they have a
• 11:06
  large order for a specific product, say product A going to
• 11:09
  a specific warehouse. They sent a massive batch, maybe
• 11:13
  a whole weeks or months worth all at once, then maybe nothing
• 11:16
  for days. OK, see where this is going?
• 11:18
  That big batched order hits the central warehouse or factory
• 11:21
  like a shockwave. They see a huge spike in demand
• 11:24
  for product A, so they scramble, ramp up production, maybe pay
• 11:27
  overtime, only for the next signal from that region to be 0
• 11:31
  for product A, but maybe now a huge spike for product B.
• 11:35
  The regional centers swing wildly between being swamped
• 11:38
  with inventory and being completely starved, and that
• 11:41
  variability gets amplified further and further upstream.
• 11:44
  It's classic sub optimization, making one small part look
• 11:47
  efficient while crippling the whole system's performance.
• 11:50
  Exactly the lean approach using Hijunka says forget optimizing
• 11:54
  individual truckloads locally, we need to establish a level
• 11:57
  predictable flow for the system. So maybe the Hijunka control
• 12:01
  dictates sending say, 4 smaller mixed load trucks every single
• 12:05
  day to that regional warehouse. The volume is level, the mix is
• 12:08
  level. Even if the trucks aren't always
• 12:10
  100% full. Even then, because that
• 12:12
  stability, that predictability, is incredibly valuable, it
• 12:15
  dramatically reduces the signal variability going upstream.
• 12:18
  And that stability allows the manufacturing plants to.
• 12:22
  Optimize globally, Precisely when the factory sees a smooth,
• 12:26
  predictable demand signal, they can stop reacting to phantom
• 12:30
  spikes and troughs. They can implement much more
• 12:32
  sophisticated scheduling, maybe moving towards producing every
• 12:36
  part every day in smaller quantities, rather than doing
• 12:39
  massive inefficient changeovers between huge batches.
• 12:42
  So leveling the downstream flow enables efficiency upstream.
• 12:46
  That's the core idea, and implementing this often requires
• 12:49
  more than just organizational changes.
• 12:51
  You need the analysis to back it up.
• 12:53
  Right, These aren't just policy decisions.
• 12:55
  You need the numbers. Absolutely.
• 12:56
  Technical modeling, simulation. They're critical.
• 13:00
  You might use simulation to combine compare different
• 13:02
  distribution strategies. You definitely use financial
• 13:05
  tools like net present value analysis to compare the total
• 13:08
  system costs of say the full truckload batching scenario
• 13:11
  versus the level flow high junk a scenario factoring in
• 13:14
  inventory cost, stock out cost overtime, everything.
• 13:17
  These quantitative tools are essential for synthesizing all
• 13:20
  the different factors logistics, finance, production and helping
• 13:24
  everyone reach consensus on the best system level design.
• 13:27
  It really strikes me how this all ties together.
• 13:29
  Whether we're talking about RTCE managing design risk early on or
• 13:33
  high junk a managing supply chain variability later, the
• 13:37
  underlying principle is this holistic systems view across the
• 13:42
  entire value stream. That's the essence of it.
• 13:44
  Product development, manufacturing, distribution.
• 13:48
  They aren't separate fiefdoms, they're interconnected parts of
• 13:51
  1 value creating system. So the goal of a truly lean
• 13:54
  enterprise is this deep integration.
• 13:57
  Yes, and if you accept that the enterprise is a complex system,
• 14:01
  then its ability to adapt and evolve successfully seems to
• 14:04
  depend on something researchers call complexity partitioning.
• 14:08
  Complexity partitioning OK, sounds intriguing.
• 14:11
  How would you explain that in maybe simpler terms for an
• 14:14
  engineer on the ground? Think about biology like the
• 14:16
  human genome. It's a master class in
• 14:18
  partitioning complexity. You have parts that are
• 14:21
  incredibly stable and conserve the core protein coding genes,
• 14:24
  the exons. They represent the reliable,
• 14:26
  proven functions. Variability there is usually bad
• 14:29
  OK. The stable core.
• 14:30
  But then you have other vast regions of DNA, historically
• 14:33
  called junk DNA, which exhibit much higher variability.
• 14:37
  These regions are now understood to be crucial for regulation,
• 14:40
  for enabling architectural recombination, for innovation,
• 14:43
  for trying out new structures without breaking the core
• 14:46
  functions. So high stability in some areas,
• 14:48
  high variability allowed in others.
• 14:50
  Exactly. A successful lean enterprise
• 14:52
  arguably needs to do the same thing.
• 14:54
  It needs to strategically decide where to enforce stability and
• 14:57
  low variability. Think standardized core
• 14:59
  processes, validated multi use components that level high junk
• 15:03
  a schedule, protect the essential value stream.
• 15:06
  But it also needs areas where flexibility, experimentation and
• 15:10
  innovation can thrive. Maybe in exploring new product
• 15:13
  features, trying radical process improvements and contain pilots
• 15:17
  or allowing more architectural freedom and non critical
• 15:19
  modules. So you partition the system to
• 15:23
  protect the stable core. Innovate at the edges.
• 15:25
  That seems to be the pattern. You need both stability for
• 15:28
  efficiency and reliability and controlled variability for
• 15:30
  adaptation and future growth. Finding that balance, that's the
• 15:34
  ongoing strategic challenge. Protecting the core value stream
• 15:37
  while enabling dynamic change elsewhere.
• 15:39
  A fascinating thought to end on until the next equation.
• 15:42
  Design Chuckies TuneIn every morning at 11.