When you're not Lean enough for JIT (Just In Time)

First off, I'm a product design engineer thinking about big quality / production concepts, so please forgive me if I'm running the buzzwords wrong. But this stuff is much more than buzzwords so I'm hoping to find guidance here.

My company makes custom assemblies of machinery. Obviously we work hard to build as much of the custom assembly from parts that are "standard" to us, but many orders contain one-off machined/welded components in a variety of stainless materials. We use a JIT ordering philosophy with everything and are applying Lean principles to the supply chain. We make none of the components ourselves - we design, purchase components, assemble, test, and ship the machinery.

Lean and JIT are being expanded and seem to make great improvements with our "standard" parts that have regular demand. We are setting up supplier agreements to stock and replenish and getting competitive pricing, higher quality, and lower overhead than before when every batch was a purchasing-managed competition.

But the one-off parts are stubborn. It is my untrained observation that orders containing these one-off parts never ship on time, usually because one of those parts has a quality or delivery issue. I can sympathize with the vendors who make these one-off parts because there is only time/money to buy the materials to make one, and our system demands perfect quality/delivery on that one-off part that to them, is essentially a prototype. Stack up a few such parts per order and the statistical likelihood of shipping on time plummets. If my vendors are only 90% on-time/in-spec on these custom parts, and I have 5 custom parts in a machine, then my probability of being ready to build on time is .90^5 = 59%, correct?

What measures might we implement to address this issue? In the near term, I can only imagine that modifying the JIT logic is required to allow some recovery opportunity.

David

Re: When you're not Lean enough for JIT

dgeesaman said:

First off, I'm a product design engineer thinking about big quality / production concepts, so please forgive me if I'm running the buzzwords wrong. But this stuff is much more than buzzwords so I'm hoping to find guidance here.

My company makes custom assemblies of machinery. Obviously we work hard to build as much of the custom assembly from parts that are "standard" to us, but many orders contain one-off machined/welded components in a variety of stainless materials. We use a JIT ordering philosophy with everything and are applying Lean principles to the supply chain. We make none of the components ourselves - we design, purchase components, assemble, test, and ship the machinery.

Lean and JIT are being expanded and seem to make great improvements with our "standard" parts that have regular demand. We are setting up supplier agreements to stock and replenish and getting competitive pricing, higher quality, and lower overhead than before when every batch was a purchasing-managed competition.

But the one-off parts are stubborn. It is my untrained observation that orders containing these one-off parts never ship on time, usually because one of those parts has a quality or delivery issue. I can sympathize with the vendors who make these one-off parts because there is only time/money to buy the materials to make one, and our system demands perfect quality/delivery on that one-off part that to them, is essentially a prototype. Stack up a few such parts per order and the statistical likelihood of shipping on time plummets. If my vendors are only 90% on-time/in-spec on these custom parts, and I have 5 custom parts in a machine, then my probability of being ready to build on time is .90^5 = 59%, correct?

What measures might we implement to address this issue? In the near term, I can only imagine that modifying the JIT logic is required to allow some recovery opportunity.

David

Click to expand...

.59% is not correct, because the delays are not in series. Vendors are independent.

You need to do a value chain analysis (bottleneck analysis) and find out from the vendor what is causing delays (late drawings, incomplete spec. late purchase orders, material availability and so on). Find the constraint and resolve it, then move on the next issue.

Re: When you're not Lean enough for JIT

David, I empathise with the situation. I have encountered similar.
There are some custom product lines and operations with there external constraints that simply will not 'fit' the internal improvement efforts and standards. In this case, appropriate internal kanban inventory may have to be established to allow for the the variation of lead time and quality deviations.
If any tool, like JIT, can not fit to the business, you make or adopt the/a tool to fit. Modification is always required somewhere. In that though, clear standards need to be set out for that. JIT requires a path of actions through the supply chain... when those can't be changed (for now), you must adapt... yes the danger is always falling to the trap of having the opposite affect... that is why I say have internal kanban and clear standard that adapts to the constraints through some math.
While that is happening, your supply chain manager, purchasers and even customer service teams need to continuously seek for vendors that can meet the qty. & quality demands, and transport improvements.
In some cases, contract review of part pricing, quality expectations and delivery times with customers may need to take place (don't forget the customer is part of the supply chain... they drive it).

Value Stream Mapping:
Map all this and when looking at costs of those things that are non-value-added and wasteful; those things that can't be changed today, seek out discussion with your customers, calculate the labour and material, space, administration of. The direct and indirect of ordering, planning, receiving, incoming inspection, storage, RMAs, shipping returns, documentation, etc. etc.

Before much more discussion, what are the specifics of your business and products?

Re: When you're not Lean enough for JIT

JIT is only worthwhile if you have good control over your processes and predictable demand. There are some cases with JIT simply does not make sense.

What you need to do is look at optimizing your costs. JIT usually optimizes costs by minimizing the costs of inventory. However, in some cases the expected costs of running out of inventory far exceed the costs of storing inventory. In this case you need to do you own optimization of costs to find the lowest cost level of inventory to maintain.

Re: When you're not Lean enough for JIT

Thank you for the quick answers.

First, the 0.9^5 = 59% is the probability of the 5 parts all being on-time and in-spec. It's been a while but I think that's correct math.

Likewise, 90% may be realistic but our goal is 98% ontime/in-spec. But that won't solve the problem - if a customer has 5 unique machines, and each machine is a little complicated and has 5 unique parts, then the probability of reaching assembly on schedule is .98^25 = 60%.

These numbers agree the orders that we struggle with.

Second, I won't be very specific but this is capital equipment used in chemical, municipal, and industrial plants. Due to the variations of each installation we adjust the length/size/style of a portion of each machine to suit. So part of the machine is made from standardized parts, and the other part is made specifically for that customer. (Once upon a time a past CEO attempted to standardize on the whole machine and use incremental sizes but it just didn't add up.)

I spoke with our purchasing director this morning and although they are in a less-than-preferred mode of operation where they are buying in advance of the automated JIT planning suggestions, he agreed that when he gets that on track the organization will expect JIT planning and lead time assumptions. He's 100% on board with my assessment that JIT alone is not a match for our entire product.

On the financial side, we as a company pay for inventory beyond the simple time value of money - there is a tax or other disincentive from our owning company that encourages us to keep things on the shelf for a minimum time and quantity. But I know it's not so severe that we have to buy everything exactly JIT. The key to identify what aspects of a component make it risky to the schedule and adjust planning accordingly.

I also earned an opportunity to present this to a task force team (set up by our CEO) that is working to improve/solve the problems with delivering orders accurately and on-time. At the moment I'm going to keep it simple and share:
- Statistically, JIT applied to order-specific parts is very risky because with our current on-time/in-spec rate of 90% for order-specific parts, it takes only 5 order-specific parts to be at 59% assembly on-schedule. And even if you push up to 98%, a block of 5 machines with 5 order-specific parts still only gets built on-time 60% of the time.
- The manufacturing reality is that some groups of parts are made to the exact quantity required and the vendor gets no "free" prototype or set-up parts. While they are similar in construction to past parts, there are still enough differences in required lead time, material sources, material specs, etc that violate the similtude / single batch process assumption. These parts will never approach 100% on-time/in-spec unless we build in something else. For example, a customer wants their part made in 2507 alloy (most of our stuff is low carbon steel, 304 S/S, or 316S/S) - and the vendor who "can do it" gets suprised by the amount of distortion in the fabrication. Suddenly the order is late because there is no time planned to replace or rework the part.

I'm not sure that anyone in our organization has assessed JIT to our parts in this way before. I know the Purchasing director had not seen it done with numbers, he simply knew that it was happening.

If there are general planning concepts that we might consider applying to this order-specific parts planning, I'm all ears.

Re: When you're not Lean enough for JIT

dgeesaman said:

If there are general planning concepts that we might consider applying to this order-specific parts planning, I'm all ears.

Click to expand...

Sounds like a routine operations research optimization problem. For a given item, what is the costs of keeping 1 on hand - 2, 3, 4, etc. Then how variable is your demand for the item? If you get low on stock, what is a good trigger point to order production of more? And can you get it produced fast enough to prevent a stock-out (none available when needed). And what is the cost if you do have a stock-out?

Through some computer brute force trial and error (Monte Carlo simulation) you can arrive at the expected cost for each item being kept in inventory, and the optimum re-order point in order to provide the lowest total expected cost.

If you google "inventory optimization" you will find a lot of vendors and techniques for doing the calculations. A wikipedia article is at http://en.wikipedia.org/wiki/Inventory_optimization

I think I need to define "order-specific" components - these are a component in a configuration that is likely to never be consumed again. These are the custom parts for each order and they cannot be forecasted.

The challenge with these parts is how to plan for them since these are always made-to-order, often the critical path, and any complications have direct impact on job schedule because no other part can be substituted.

David

David - if these 1 off's are critical but you have issues in quality/ delivery, may I suggest costing out the impact of this.

Your company may be penny wise and dollar foolish in how you select and manage your vendors. I don't know how attractive your business is to your suppliers but it may be a nuisance and not lucrative / repeatable enough for them to care.

You may want to be offering a 10% bonus above normal if it meets quality + delivery performance.

dgeesaman said:

I think I need to define "order-specific" components - these are a component in a configuration that is likely to never be consumed again. These are the custom parts for each order and they cannot be forecasted.

The challenge with these parts is how to plan for them since these are always made-to-order, often the critical path, and any complications have direct impact on job schedule because no other part can be substituted.

David

Click to expand...

Still it is an issue of optimizing costs. If the cost of being late on an item was SO expensive, it would be worth making two fo the critical items and scrap one if both are okay. Or, it may be worth extra "expediting" efforts on those critical parts. Even on "one of a kind" items there are precautions that can be taken, if the alternative is painful/costly enough. Might want to look at some other industries that basically supply custom-made parts for ideas.

Still it is an issue of optimizing costs. If the cost of being late on an item was SO expensive, it would be worth making two fo the critical items and scrap one if both are okay. Or, it may be worth extra "expediting" efforts on those critical parts. Even on "one of a kind" items there are precautions that can be taken, if the alternative is painful/costly enough. Might want to look at some other industries that basically supply custom-made parts for ideas.

Click to expand...

Considering that these one-of-a-kind parts are some of the physically largest and most expensive parts on the order, I would construct a bulletproof glass ceiling if I suggested buying two and scrapping the other. (It also doesn't address the issue of underestimated manufacturing time). But your point is taken - I think the first thing our organization need to do is understand which parts are "one-of-a-kind" and what risks are involved. Then implement the measures that make the most sense, or even bring in expertise to assist. We were trained to use our ERP system in a simple lead time / JIT planning approach and we'll be looking into some unused features to help manage the not-so-JIT parts. We do so many such parts, and they look simple enough, that there is a complacency about it in spite of the data and our experience.

My brief research led me to OKP as a possible strategy. I haven't bought the book or been trained in this philosophy, but on the surface it describes my organization's product well.

Thanks for the help. It was a struggle to find information describing where JIT is not entirely appropriate, as the Internet is littered with overly-simplified articles about the unqualified wonderfulness of JIT.

David