You are on page 1of 5

Little’s Law and WIP

Please see the excel file which shows the CT and MLT under two scenarios:

1. Unsynchronized flow: Where station A starts working on a new part as soon as it


finishes with the previous one. See how MLT keeps on increasing over here while the
CT is still 26 minutes.

2. Synchronized flow: Where station A works for 14 minutes then sits idle for 12 minutes
and then works again on parts (increment of 26 mins). Here again the CT is 26 minutes
but MLT is 76 (Note that when there is no WIP inventory a part WILL come out in
76 minutes which is speed of A (14) + B(13) + D(25)+ E(24).... Note again how
processing time of C is not counted here as in the synchronous mode C and B will finish
at the exact same time. A will now start in such a way that the part comes out of B at
the exact same time as C.)

Here is another interesting thought: When there is no traffic jam the lowest possible MLT should
ideally be the equal to the longest sequence of dependent events (considering resource constraints).

You might also see something else interesting: No matter how fast A starts working, the finish
times of the corresponding units (i.e. the Operation E finish time) is exactly the same under the
two scenarios (See column L). Well, then you would say what's the point of slowing A down?
Instead of starting late at A we might as well start as fast as possible and let it accumulate as
WIP? (Answer: Don't be myopic! See what is the lead time you will quote for a customized
order which arrives at say 308 minutes under each of the two scenarios? )

1. Under the unsynchronized scenario: the WIP in the system at time 308 = 308/14 - 308/26
=10.15
MLT at time 308 = 10.15 * 26 =264 minutes.

How much will you quote for the order that comes at time 308? = Another 264 minutes to clear
the existing 10.15 parts (known as MLT) + 26 minutes for processing the new order (at C) +
49 minutes for new order (at D & E) = 339 minutes approx. (Note here how due to
synchronization, the processing times of A +B are not counted separately)

Quoted time for an order received at time 308 under the first scenario = 308+ 339 = 647 minutes
Note that when we solve problems in OM we generally ignore the time entries written in red
as the touch times are too small compared to the MLTs. We simply quote MLT= 264 minutes
which is the time required to clear the existing 10.15 parts and generally ignore the processing
requirements of the new order itself (at the bottleneck and the subsequent stations). If the WIP
is very large say 100 units then the Quoted time will be quite close to the MLT.

Also note that MLT is the function of WIP!! It changes as the WIP changes.

Scenario 1: Unsynchronized Flow

A (start A(Finish B(Start B(Finish C(Start C(Finish D(Start D(Finish E(Start E(Finish
Unit# CT MLT
time) Time) Time) Time) time) Time) time) Time) time) Time)

1 0 14 14 27 0 26 27 52 52 76 76
2 14 28 28 41 26 52 52 77 77 101 25 87
3 28 42 42 55 52 78 78 103 103 127 26 99
4 42 56 56 69 78 104 104 129 129 153 26 111
5 56 70 70 83 104 130 130 155 155 179 26 123
6 70 84 84 97 130 156 156 181 181 205 26 135
7 84 98 98 111 156 182 182 207 207 231 26 147
8 98 112 112 125 182 208 208 233 233 257 26 159
9 112 126 126 139 208 234 234 259 259 283 26 171
10 126 140 140 153 234 260 260 285 285 309 26 183
11 140 154 154 167 260 286 286 311 311 335 26 195
12 154 168 168 181 286 312 312 337 337 361 26 207
13 168 182 182 195 312 338 338 363 363 387 26 219
14 182 196 196 209 338 364 364 389 389 413 26 231
15 196 210 210 223 364 390 390 415 415 439 26 243
16 210 224 224 237 390 416 416 441 441 465 26 255
17 224 238 238 251 416 442 442 467 467 491 26 267
18 238 252 252 265 442 468 468 493 493 517 26 279
19 252 266 266 279 468 494 494 519 519 543 26 291
20 266 280 280 293 494 520 520 545 545 569 26 303
21 280 294 294 307 520 546 546 571 571 595 26 315
22 294 308 308 321 546 572 572 597 597 621 26 327
23 308 322 322 335 572 598 598 623 623 647 26 339
24 322 336 336 349 598 624 624 649 649 673 26 351
25 336 350 350 363 624 650 650 675 675 699 26 363
26 350 364 364 377 650 676 676 701 701 725 26 375
27 364 378 378 391 676 702 702 727 727 751 26 387
28 378 392 392 405 702 728 728 753 753 777 26 399
29 392 406 406 419 728 754 754 779 779 803 26 411
30 406 420 420 433 754 780 780 805 805 829 26 423
2. Under scenario 2 you would have simply quoted 308 + 76 minutes as the delivery time.

Scenario 2: Synchronized Flow

A (start A(Finish B(Start B(Finish C(Start C(Finish D(Start D(Finish E(Start E(Finish
Unit# CT MLT
time) Time) Time) Time) time) Time) time) Time) time) Time)

1 0 14 14 27 0 26 27 52 52 76 76

2 25 39 39 52 26 52 52 77 77 101 25 76

3 51 65 65 78 52 78 78 103 103 127 26 76

4 77 91 91 104 78 104 104 129 129 153 26 76

5 103 117 117 130 104 130 130 155 155 179 26 76

6 129 143 143 156 130 156 156 181 181 205 26 76

7 155 169 169 182 156 182 182 207 207 231 26 76

8 181 195 195 208 182 208 208 233 233 257 26 76

9 207 221 221 234 208 234 234 259 259 283 26 76

10 233 247 247 260 234 260 260 285 285 309 26 76

11 259 273 273 286 260 286 286 311 311 335 26 76

12 285 299 299 312 286 312 312 337 337 361 26 76

13 311 325 325 338 312 338 338 363 363 387 26 76

14 337 351 351 364 338 364 364 389 389 413 26 76

15 363 377 377 390 364 390 390 415 415 439 26 76

16 389 403 403 416 390 416 416 441 441 465 26 76

17 415 429 429 442 416 442 442 467 467 491 26 76

18 441 455 455 468 442 468 468 493 493 517 26 76

19 467 481 481 494 468 494 494 519 519 543 26 76

20 493 507 507 520 494 520 520 545 545 569 26 76

21 519 533 533 546 520 546 546 571 571 595 26 76

22 545 559 559 572 546 572 572 597 597 621 26 76

23 571 585 585 598 572 598 598 623 623 647 26 76

24 597 611 611 624 598 624 624 649 649 673 26 76

25 623 637 637 650 624 650 650 675 675 699 26 76

26 649 663 663 676 650 676 676 701 701 725 26 76

27 675 689 689 702 676 702 702 727 727 751 26 76

28 701 715 715 728 702 728 728 753 753 777 26 76

29 727 741 741 754 728 754 754 779 779 803 26 76

30 753 767 767 780 754 780 780 805 805 829 26 76
So the benefits of delaying the start of A (read non-bottleneck) are ginormous:

First of all no WIP build-up which means less crowding, less storage requirements and better
information flow. Also, when you have no/little WIP, the MLT also does not fluctuate much
(see MLT variation in first scenario) and hence it becomes easy to estimate and meet the
deadlines to your customers. Slowing A down also means less raw materials consumed.
Sometimes you might run out of raw materials to execute the new customized order itself if
you are running A at full efficiency. Then you might have to procure raw materials from a
supplier in between and it might take its own sweet time and could be super expensive. Now
imagine if your supplier operates the same way you do (obsessing with efficiency everywhere)
then her MLT is going to be much longer than just the transportation time. So the replenishment
time from your supplier is her own production time plus the transportation time. The latter is
generally a tiny portion of the actual replenishment time. You must have seen this: trucks take
maximum 2 days to reach your plant from the supplier while she asks for 1 month to supply
your order itself. A shoe that should take 1 hour takes 2-3 whole months to reach from
Bangladesh to USA where the ship travel time is only 15 days! The touch time for a shoe as a
percentage of the total time it requires to reach to the end consumer, which is about 6 months,
is so insignificant. The entire time the shoe is actually waiting at different places. Geographical
proximity is not same as being close to the customer. Your customer might be right outside
your factory doors but you might still be 6 months away in delivering her orders.).

Look at the other side now: Imagine if it takes one month for your raw material supplier to
supply your orders how much raw material inventory will you keep in your r.m. warehouses?
At least 2-3 months’ worth, to ensure availability and to take benefits of quantity discounts?
Now can you predict the consumption of the 10000 different varieties of raw materials
properly? You will realize you will have to "forecast" the raw material consumption needs for
all the raw materials if you don't want to wait the way Donner did every time it got customized
orders. Also you will need computers to do your materials requirement planning (known as
MRP). And when you forecast, no matter how well you do it, you WILL run short of some
items and run excess on others despite the "best forecasting software" that exists in the market?
Please realize a car cannot be delivered even if one screw is missing! So even if one part is
missing the customized order cannot be complete. Now even if you are 99% accurate on one
raw material, thanks to the "better forecasts", what is the probability that a car which requires
10000 different items will be delivered in time? Not to mention your inventories are going
through the roof if you want to be more and more accurate on your inventories. You have huge
RM inventory, huge WIP and God forbid if the market is the constraint then even huge FG
inventory. It is insane amount of cash blocked in inventory. We then try and give "discounts"
to our customers and "incentives" to our salespeople to "push" the inventories to the next person
in the supply chain. And when you give discounts on large quantities each and every one of
your customer WILL HAVE to buy in large quantities (can you take a guess why?) even when
they don't want to or can't afford to. Now the customer is struggling with cash problems as her
money is blocked in inventory of some limited variety of items, which by the way, in many
cases, is large enough for her grandchildren. Who do you think is funding all the inventory for
everyone in the supply chain (your/ supplier's/customer's)? BANKS! At the end of the day
everyone will tell you they all work for the banks! Each and every one is struggling with CASH
problems, profit problems, ROI problems. This story does not end here and will make your
heads spin when you start seeing what sort of decisions are companies making to increase their
profits instead of looking at the real issue. Examples: laying off workers when the direct labor
costs are nowhere near the inventory costs, pushing suppliers to achieve year on year reduction
in raw materials costs, buying ERP systems to "improve information flow" and "reduce the
forecasting inaccuracies", incentivizing customers to dig their own graves by offering more
discounts and higher credit terms, buying faster machines to save 2 seconds on non-bottlenecks
(thus further adding to the cash woes both while buying the machine as well then creating WIP
at faster rates than earlier), taking up cost reduction programs in every part of the company that
is imaginable (plastic plates, L1 tenders), etc. This is carpet bombing in its truest sense: When
you don't know where the enemy is you will bomb everywhere and everyone. The worst part
is you won't even know which bomb killed the enemy (if it did in the first place) and if the
money spent on killing the enemy was not higher than the benefits reaped. And so year on year
you keep repeating the same mistakes as you don't know what really works....You can build
the whole cause and effect like this and there is no end to the damage that not knowing a
bottleneck can do to you.

So what do we learn?

Slow the heck down at every stage, unless you are at the bottleneck! Efficiency matters only
at one and only one place!

How many bottlenecks do you generally expect in a real life system? Refer to all the cases you
have gone through and see how many.

You might also like