1 00:00:00,000 --> 00:00:02,430 The following content is provided under a Creative 2 00:00:02,430 --> 00:00:03,730 Commons License. 3 00:00:03,730 --> 00:00:06,060 Your support will help MIT OpenCourseWare 4 00:00:06,060 --> 00:00:10,090 continue to offer high quality educational resources for free. 5 00:00:10,090 --> 00:00:12,690 To make a donation or to view additional materials 6 00:00:12,690 --> 00:00:16,560 from hundreds of MIT courses, visit MIT OpenCourseWare 7 00:00:16,560 --> 00:00:17,744 at ocw.mit.edu. 8 00:00:25,790 --> 00:00:28,120 PROFESSOR: So welcome to Lean Thinking Part II. 9 00:00:28,120 --> 00:00:29,840 We're essentially going to pick up 10 00:00:29,840 --> 00:00:32,930 where we left off last time in discussing 11 00:00:32,930 --> 00:00:38,540 the Lean framework of thinking about value, the value stream, 12 00:00:38,540 --> 00:00:41,370 flow, pull, and perfection. 13 00:00:41,370 --> 00:00:45,830 Hopefully the video gave you some real active sort 14 00:00:45,830 --> 00:00:48,500 of visual representations of what some of these concepts 15 00:00:48,500 --> 00:00:50,280 might mean. 16 00:00:50,280 --> 00:00:52,490 So let's talk about flow. 17 00:00:52,490 --> 00:00:54,258 We would like, if we have a value stream-- 18 00:00:54,258 --> 00:00:55,550 this is a patient value stream. 19 00:00:55,550 --> 00:00:57,380 We have people coming in and going 20 00:00:57,380 --> 00:01:00,800 through diagnostic procedures, and hopefully leaving in better 21 00:01:00,800 --> 00:01:02,240 shape than they came in. 22 00:01:02,240 --> 00:01:05,152 And we would like to create flow in this system. 23 00:01:05,152 --> 00:01:06,860 We have to think about a bunch of things. 24 00:01:06,860 --> 00:01:09,410 We obviously want to focus on what is flowing. 25 00:01:09,410 --> 00:01:15,090 Again, the value stream map is a map that looks at the product, 26 00:01:15,090 --> 00:01:20,910 in this case, the patient flowing through the process. 27 00:01:20,910 --> 00:01:24,450 We don't want to be limited by organizational or people 28 00:01:24,450 --> 00:01:24,950 boundaries. 29 00:01:24,950 --> 00:01:28,490 It's not a value stream map if it follows organizational work 30 00:01:28,490 --> 00:01:29,990 or an individual's work. 31 00:01:29,990 --> 00:01:33,650 It's really about the product. 32 00:01:33,650 --> 00:01:35,600 And first order, we want to make sure 33 00:01:35,600 --> 00:01:40,010 that there's the minimum amount of buffers in there, inventory. 34 00:01:40,010 --> 00:01:41,630 Here's a waiting room full of people. 35 00:01:41,630 --> 00:01:43,460 That's not flow, right? 36 00:01:43,460 --> 00:01:47,090 And it's interesting to me, as a non-medical professional, 37 00:01:47,090 --> 00:01:52,160 to just sort of observe things from a Lean perspective. 38 00:01:52,160 --> 00:01:56,690 I have had both good and bad experiences as a patient, 39 00:01:56,690 --> 00:01:58,190 or the parent of patients. 40 00:01:58,190 --> 00:01:59,400 Sometimes it's this. 41 00:01:59,400 --> 00:02:03,370 You're sitting in the waiting room for five or six hours. 42 00:02:03,370 --> 00:02:06,180 And sometimes it isn't. 43 00:02:06,180 --> 00:02:09,360 I've had sort of astonishing experiences 44 00:02:09,360 --> 00:02:13,980 with well-organized medical facilities, where 45 00:02:13,980 --> 00:02:15,990 literally once, I didn't have time to sit down. 46 00:02:15,990 --> 00:02:17,670 They gave me a little form to fill out, 47 00:02:17,670 --> 00:02:20,100 and before I had finished filling out the form, 48 00:02:20,100 --> 00:02:22,290 they were calling my name to see the doctor. 49 00:02:22,290 --> 00:02:25,920 It was it was a well-organized system that had flow. 50 00:02:25,920 --> 00:02:28,350 But bottlenecks are going to slow that down. 51 00:02:28,350 --> 00:02:30,780 Buffers, essentially the same. 52 00:02:30,780 --> 00:02:33,270 Wait time in between steps. 53 00:02:33,270 --> 00:02:34,560 Sometimes that's necessary. 54 00:02:34,560 --> 00:02:37,230 You have an expensive facility like a CAT scanner. 55 00:02:37,230 --> 00:02:40,238 You're not going to risk not using that facility, 56 00:02:40,238 --> 00:02:42,030 so people are going to have to wait for it. 57 00:02:42,030 --> 00:02:44,430 But to achieve flow, you need to minimize that. 58 00:02:44,430 --> 00:02:46,980 And one of the key things to think about 59 00:02:46,980 --> 00:02:50,290 is time it's an essential metric. 60 00:02:50,290 --> 00:02:52,270 Our discussion today is going to really 61 00:02:52,270 --> 00:02:56,440 be about how do we measure and define time. 62 00:02:56,440 --> 00:02:58,180 Time is sort of an obvious thing. 63 00:02:58,180 --> 00:03:01,030 But when you start thinking about measuring and defining 64 00:03:01,030 --> 00:03:03,000 it, it does get a little bit complicated. 65 00:03:03,000 --> 00:03:06,730 You have to think about it carefully. 66 00:03:06,730 --> 00:03:10,420 And the key is really to understand the definition 67 00:03:10,420 --> 00:03:12,830 of how the time is measured. 68 00:03:12,830 --> 00:03:15,970 And one of the problems here is that the terminology-- 69 00:03:15,970 --> 00:03:18,010 there's some terminology here-- 70 00:03:18,010 --> 00:03:19,190 is not standardized. 71 00:03:19,190 --> 00:03:22,870 So in different fields, industrial engineers and supply 72 00:03:22,870 --> 00:03:25,450 chain people will talk about things using the same words. 73 00:03:25,450 --> 00:03:27,310 They mean different things. 74 00:03:27,310 --> 00:03:30,190 Sometimes the words are used in a way that isn't intuitive. 75 00:03:30,190 --> 00:03:32,200 So rather than try to teach you the jargon 76 00:03:32,200 --> 00:03:35,050 of any particular field, what we're going to ask you to do 77 00:03:35,050 --> 00:03:39,280 is think about time and make sure you understand 78 00:03:39,280 --> 00:03:41,620 and communicate your definitions of time. 79 00:03:41,620 --> 00:03:43,870 And we'll actually have a little exercise about that. 80 00:03:43,870 --> 00:03:44,710 So wait time. 81 00:03:44,710 --> 00:03:46,840 That's relatively simple. 82 00:03:46,840 --> 00:03:49,660 If something is just sitting there, it is waiting. 83 00:03:49,660 --> 00:03:52,460 So there is work in progress. 84 00:03:52,460 --> 00:03:55,960 The work that you're trying to do is not moving forward. 85 00:03:55,960 --> 00:03:56,980 That's wait time. 86 00:03:56,980 --> 00:03:59,140 And it's also called queue time, delay time. 87 00:03:59,140 --> 00:04:01,630 Again, the terminology isn't particularly standard. 88 00:04:01,630 --> 00:04:03,737 The concept is pretty easy to understand, though. 89 00:04:03,737 --> 00:04:06,070 Processing time, on the other hand, is a little bit more 90 00:04:06,070 --> 00:04:07,670 complicated. 91 00:04:07,670 --> 00:04:10,990 So now something is happening to the work product. 92 00:04:10,990 --> 00:04:13,180 Some sort of resources are being consumed. 93 00:04:13,180 --> 00:04:14,830 Maybe a person is working on it. 94 00:04:14,830 --> 00:04:16,570 Maybe a machine is working on it. 95 00:04:16,570 --> 00:04:19,180 And that could be value added or non-value added. 96 00:04:19,180 --> 00:04:23,950 It could be an active or relatively passive use. 97 00:04:23,950 --> 00:04:25,510 There are multiple names for it. 98 00:04:25,510 --> 00:04:29,860 Touch time-- I kind of like touch time 99 00:04:29,860 --> 00:04:32,740 because it's intuitive-- means, in, 100 00:04:32,740 --> 00:04:35,440 for example, engineering or health care work-- 101 00:04:35,440 --> 00:04:37,780 someone is working on it. 102 00:04:37,780 --> 00:04:40,150 Both of those fields are very people intensive. 103 00:04:40,150 --> 00:04:42,160 If somebody is working with the patient 104 00:04:42,160 --> 00:04:45,400 or working with the diagnosis or working with the engineering 105 00:04:45,400 --> 00:04:47,890 drawing, that's touch time. 106 00:04:47,890 --> 00:04:51,100 It's also sometimes called in process time or response 107 00:04:51,100 --> 00:04:54,730 time in the sense of how quickly can the people respond 108 00:04:54,730 --> 00:04:58,330 to a demand that they get something done. 109 00:04:58,330 --> 00:05:00,692 And these things tend to pile up in ways that 110 00:05:00,692 --> 00:05:01,900 are kind of illustrated here. 111 00:05:01,900 --> 00:05:05,120 In a simple case, if you had some work you need to get done, 112 00:05:05,120 --> 00:05:07,370 well, first it waits. 113 00:05:07,370 --> 00:05:08,010 Question. 114 00:05:08,010 --> 00:05:11,270 AUDIENCE: [INAUDIBLE] there was an oven [INAUDIBLE] 115 00:05:11,270 --> 00:05:14,072 shoes, and you heat the thing up, 116 00:05:14,072 --> 00:05:16,130 but then you would open the oven to let it sit 117 00:05:16,130 --> 00:05:18,200 and not touch it because it was cooling off. 118 00:05:18,200 --> 00:05:20,058 PROFESSOR: That's right. 119 00:05:20,058 --> 00:05:20,600 That's right. 120 00:05:20,600 --> 00:05:21,560 That's classic, right? 121 00:05:21,560 --> 00:05:25,260 OK, so is that wait time because it's cooling off? 122 00:05:25,260 --> 00:05:27,060 But it's occupying the oven. 123 00:05:27,060 --> 00:05:28,272 It's not touch time. 124 00:05:28,272 --> 00:05:29,480 The guy is not working on it. 125 00:05:29,480 --> 00:05:31,400 He can go do something else. 126 00:05:31,400 --> 00:05:33,860 But the equipment is used. 127 00:05:33,860 --> 00:05:34,910 So that's classic. 128 00:05:34,910 --> 00:05:36,740 And I don't have an easy answer for you. 129 00:05:36,740 --> 00:05:38,240 It depends on what you're measuring. 130 00:05:38,240 --> 00:05:40,370 If you're measuring the equipment, 131 00:05:40,370 --> 00:05:41,510 how many ovens do we need? 132 00:05:41,510 --> 00:05:43,680 You better account for that cooling 133 00:05:43,680 --> 00:05:45,680 because they probably don't want to yank it out. 134 00:05:45,680 --> 00:05:46,580 It'll cool too fast. 135 00:05:49,220 --> 00:05:52,550 So if your concern is the oven, you got to count that. 136 00:05:52,550 --> 00:05:54,860 If your concern is the worker, that part doesn't count. 137 00:05:54,860 --> 00:05:59,450 If your concern is the shoe, it's in process in some way. 138 00:05:59,450 --> 00:06:01,610 It's cooling off. 139 00:06:01,610 --> 00:06:03,890 You can't say that's wait time because you 140 00:06:03,890 --> 00:06:06,320 can't make that go faster. 141 00:06:06,320 --> 00:06:08,450 Another classic one is painting. 142 00:06:08,450 --> 00:06:10,010 You can't make paint dry faster. 143 00:06:10,010 --> 00:06:10,640 Well, you can. 144 00:06:10,640 --> 00:06:13,250 You can blast hot air on it, but that may or may not 145 00:06:13,250 --> 00:06:15,920 be a good thing. 146 00:06:15,920 --> 00:06:18,350 And that's exactly what I'm trying to get across here, 147 00:06:18,350 --> 00:06:23,150 is that the time has a lot of sort of unexpected complexity 148 00:06:23,150 --> 00:06:24,860 when you're looking at these things, 149 00:06:24,860 --> 00:06:27,030 and you have to think pretty straight about it. 150 00:06:27,030 --> 00:06:29,150 So you could be waiting, you could 151 00:06:29,150 --> 00:06:30,650 be doing value added work, you could 152 00:06:30,650 --> 00:06:33,560 be non-value added, but still in processing. 153 00:06:33,560 --> 00:06:35,270 We stuck the glue on. 154 00:06:35,270 --> 00:06:36,770 Now we're cooling off. 155 00:06:36,770 --> 00:06:39,170 We're still using the oven. 156 00:06:39,170 --> 00:06:41,590 We can't leave the oven, but it's not really value added. 157 00:06:41,590 --> 00:06:43,340 Maybe we could do that in a different way. 158 00:06:43,340 --> 00:06:45,740 Maybe it's necessary non-value added. 159 00:06:45,740 --> 00:06:48,110 Maybe we'll call that yellow. 160 00:06:48,110 --> 00:06:49,640 At some point the processing ends 161 00:06:49,640 --> 00:06:52,310 and we go back into inventory, and we're waiting again. 162 00:06:52,310 --> 00:06:55,400 Sort of imagine that cycle repeated lots of times 163 00:06:55,400 --> 00:06:57,770 to finish our overall process. 164 00:06:57,770 --> 00:07:01,280 And cycle time is our overall time. 165 00:07:01,280 --> 00:07:04,160 There isn't a whole lot of confusion about cycle time 166 00:07:04,160 --> 00:07:08,210 means overall time, wait plus processing. 167 00:07:08,210 --> 00:07:09,928 There's some other names for it. 168 00:07:09,928 --> 00:07:11,720 The thing you have to really watch out for, 169 00:07:11,720 --> 00:07:14,690 cycle time is cycle of what? 170 00:07:14,690 --> 00:07:16,220 It could be the cycle of a machine, 171 00:07:16,220 --> 00:07:17,900 ka-chunk, ka-chunk, ka-chunk. 172 00:07:17,900 --> 00:07:19,250 That's seconds. 173 00:07:19,250 --> 00:07:22,070 It could be cycle of the overall shoe manufacturing 174 00:07:22,070 --> 00:07:25,190 process, which, in New Balance, was an amazingly short 175 00:07:25,190 --> 00:07:26,810 three days. 176 00:07:26,810 --> 00:07:30,140 But in aeronautical, engineering, cycle time 177 00:07:30,140 --> 00:07:33,083 of an aircraft life cycle, 20 years. 178 00:07:33,083 --> 00:07:35,000 Cycle time can mean a lot of different things. 179 00:07:35,000 --> 00:07:38,750 You've got to just define your boundaries very carefully 180 00:07:38,750 --> 00:07:41,340 when you're talking about cycle time. 181 00:07:41,340 --> 00:07:42,860 So now we have a little exercise. 182 00:07:42,860 --> 00:07:45,800 We're going to ask you to tackle some of these complexities. 183 00:07:45,800 --> 00:07:49,280 I hope everybody held on to this sheet. 184 00:07:49,280 --> 00:07:51,980 You were using this before. 185 00:07:51,980 --> 00:07:54,170 We call it 5Sing out of existence. 186 00:07:54,170 --> 00:07:56,550 You clean up something you really need, 187 00:07:56,550 --> 00:07:57,900 and then it's gone. 188 00:07:57,900 --> 00:08:01,730 I hope that didn't happen because now we're 189 00:08:01,730 --> 00:08:03,590 going to look at the times, and we're 190 00:08:03,590 --> 00:08:05,160 going to do a little exercise. 191 00:08:05,160 --> 00:08:08,150 We're going to add up the overall cycle time of an order 192 00:08:08,150 --> 00:08:09,800 from the time the customer walks up 193 00:08:09,800 --> 00:08:11,660 to the time they get their hot dog. 194 00:08:11,660 --> 00:08:14,090 The tricky thing is these times are not 195 00:08:14,090 --> 00:08:16,460 necessarily in the right units. 196 00:08:16,460 --> 00:08:19,910 Make sure everything is converted to time per order. 197 00:08:19,910 --> 00:08:21,800 Is it time for a hot dog? 198 00:08:21,800 --> 00:08:22,760 Subtle. 199 00:08:22,760 --> 00:08:23,960 What's the average order? 200 00:08:23,960 --> 00:08:26,660 Does anybody remember? 201 00:08:26,660 --> 00:08:29,510 Two hot dogs, right? 202 00:08:29,510 --> 00:08:31,190 And we got to worry about rework. 203 00:08:31,190 --> 00:08:33,120 There's some percentage chance- we're 204 00:08:33,120 --> 00:08:35,120 looking at the average-- there's some percentage 205 00:08:35,120 --> 00:08:38,630 chance that things are going to have to be done over. 206 00:08:38,630 --> 00:08:41,750 So this actually gets pretty well 207 00:08:41,750 --> 00:08:44,490 at how tricky this can sometimes be. 208 00:08:44,490 --> 00:08:44,990 so? 209 00:08:44,990 --> 00:08:48,530 Try to normalize the times, sum them up, and use the sheet. 210 00:08:48,530 --> 00:08:50,450 This is actually a group exercise. 211 00:08:50,450 --> 00:08:52,280 If you guys want to divide and conquer, 212 00:08:52,280 --> 00:08:55,760 or if you just want to see what people get, 213 00:08:55,760 --> 00:09:00,560 whatever, whatever strategy you want to take to puzzle 214 00:09:00,560 --> 00:09:01,370 this out is fine. 215 00:09:01,370 --> 00:09:06,410 But you only have a little under 10 minutes. 216 00:09:06,410 --> 00:09:08,180 Why don't we give it till, like, three 217 00:09:08,180 --> 00:09:09,950 after to do this exercise? 218 00:09:13,310 --> 00:09:16,040 AUDIENCE: I see what you're saying. 219 00:09:16,040 --> 00:09:18,860 It shouldn't take 20 seconds to put a hot dog in a bun. 220 00:09:18,860 --> 00:09:23,930 So 20 seconds to put it in the bun, and then about 10 seconds 221 00:09:23,930 --> 00:09:25,550 putting it [INAUDIBLE]. 222 00:09:25,550 --> 00:09:26,600 AUDIENCE: Exactly. 223 00:09:26,600 --> 00:09:28,130 That's how I interpret it. 224 00:09:28,130 --> 00:09:31,400 PROFESSOR: All righty, what do we got? 225 00:09:31,400 --> 00:09:33,950 One of the fun things about this class 226 00:09:33,950 --> 00:09:41,280 is our sort of unintentional experimenting and work styles. 227 00:09:41,280 --> 00:09:43,550 Some people were just doing away. 228 00:09:43,550 --> 00:09:46,880 Other people were talking waving their arms and stuff. 229 00:09:46,880 --> 00:09:48,570 It was interesting. 230 00:09:48,570 --> 00:09:52,760 Let's see if what this process yielded. 231 00:09:52,760 --> 00:10:00,150 I'm numbering you guys M1-3 and E1 and 2. 232 00:10:00,150 --> 00:10:04,100 So first medical option table. 233 00:10:04,100 --> 00:10:06,030 What do you got? 234 00:10:06,030 --> 00:10:07,620 449. 235 00:10:07,620 --> 00:10:09,180 OK. 236 00:10:09,180 --> 00:10:10,560 Interesting. 237 00:10:10,560 --> 00:10:12,660 M2. 238 00:10:12,660 --> 00:10:19,440 393. 239 00:10:19,440 --> 00:10:19,940 OK. 240 00:10:23,680 --> 00:10:24,850 What do you guys got? 241 00:10:24,850 --> 00:10:26,330 You have different things. 242 00:10:26,330 --> 00:10:27,710 Give me a sample. 243 00:10:27,710 --> 00:10:30,537 See, this is the table that didn't talk. 244 00:10:30,537 --> 00:10:32,000 AUDIENCE: [INAUDIBLE] around 300. 245 00:10:32,000 --> 00:10:33,200 PROFESSOR: Around 300. 246 00:10:33,200 --> 00:10:34,490 OK. 247 00:10:34,490 --> 00:10:36,650 Plus or minus. 248 00:10:36,650 --> 00:10:37,220 All right. 249 00:10:37,220 --> 00:10:39,560 And E2. 250 00:10:39,560 --> 00:10:40,820 432. 251 00:10:40,820 --> 00:10:41,600 OK. 252 00:10:41,600 --> 00:10:42,260 There we go. 253 00:10:44,820 --> 00:10:45,470 There we go. 254 00:10:49,480 --> 00:10:51,720 So we didn't exactly agree. 255 00:10:51,720 --> 00:10:54,300 We're going to give you guys sort of an answer sheet 256 00:10:54,300 --> 00:10:56,820 later so you can figure out. 257 00:10:56,820 --> 00:11:00,960 But it sort of illustrates the issue 258 00:11:00,960 --> 00:11:03,240 of the squirminess of this. 259 00:11:03,240 --> 00:11:06,480 It's a very simple process. 260 00:11:06,480 --> 00:11:07,897 We gave you the times. 261 00:11:07,897 --> 00:11:09,480 But how do you interpret them exactly? 262 00:11:09,480 --> 00:11:10,522 How do you do the rework? 263 00:11:10,522 --> 00:11:12,300 Do you average that in? 264 00:11:12,300 --> 00:11:14,190 Probably the most reasonable thing to do 265 00:11:14,190 --> 00:11:18,030 would be to add 10% to all of the things that 266 00:11:18,030 --> 00:11:20,550 might have to be done over. 267 00:11:20,550 --> 00:11:22,740 But there's actually a couple of do over loops, 268 00:11:22,740 --> 00:11:25,170 so do you add 20% to that one? 269 00:11:25,170 --> 00:11:25,980 I'm not sure. 270 00:11:25,980 --> 00:11:28,440 You could interpret that a couple of different ways. 271 00:11:28,440 --> 00:11:32,940 More importantly, or more significantly-- 272 00:11:32,940 --> 00:11:34,770 it has a bigger impact on the numbers-- 273 00:11:34,770 --> 00:11:36,360 what do you do with the stuff that's 274 00:11:36,360 --> 00:11:39,570 kind of off the main process loop there, the cleanup? 275 00:11:39,570 --> 00:11:43,590 Do e you average that and put it in as a sort of tax 276 00:11:43,590 --> 00:11:45,600 against each hot dog order? 277 00:11:45,600 --> 00:11:48,060 Or do you say nah, they'll do that when it's quiet. 278 00:11:48,060 --> 00:11:49,950 It shouldn't count. 279 00:11:49,950 --> 00:11:52,320 And that's probably the difference between the people 280 00:11:52,320 --> 00:11:56,520 that we're up above 400 and the people down around 300. 281 00:11:56,520 --> 00:11:58,210 Does that make sense? 282 00:11:58,210 --> 00:12:00,150 I have the 300 people were leaving that out 283 00:12:00,150 --> 00:12:05,220 because otherwise it's hard to get a number that small. 284 00:12:05,220 --> 00:12:10,170 This is pretty much-- 285 00:12:10,170 --> 00:12:13,500 that's somehow one second higher than the number 286 00:12:13,500 --> 00:12:17,430 you get if you add 10% to everything and tax 287 00:12:17,430 --> 00:12:19,350 with the cleanup. 288 00:12:19,350 --> 00:12:21,840 So that's real close to that number. 289 00:12:21,840 --> 00:12:24,750 So there's different ways to do it, and they're all legitimate. 290 00:12:24,750 --> 00:12:28,920 But it gives you a feel for how this is not an exact science. 291 00:12:28,920 --> 00:12:30,900 And again, if we wanted to get more exact, 292 00:12:30,900 --> 00:12:32,160 we would have to go to the [INAUDIBLE].. 293 00:12:32,160 --> 00:12:33,720 We might have to stand there with a stopwatch 294 00:12:33,720 --> 00:12:35,220 and watch the customers come and go, 295 00:12:35,220 --> 00:12:38,790 and see what actually does happen. 296 00:12:38,790 --> 00:12:41,733 For example, the cleanup issue. 297 00:12:41,733 --> 00:12:43,650 Do they really have time to do it on the side, 298 00:12:43,650 --> 00:12:45,480 or do they have to actually make customers 299 00:12:45,480 --> 00:12:46,590 wait while they do that? 300 00:12:46,590 --> 00:12:48,330 You don't know until you go look. 301 00:12:48,330 --> 00:12:54,812 So again, an illustration in how this can be a little tricky. 302 00:12:54,812 --> 00:12:56,520 And we're going to go even deeper on this 303 00:12:56,520 --> 00:12:57,670 later this afternoon. 304 00:12:57,670 --> 00:13:01,150 But for now, that exercise is done. 305 00:13:01,150 --> 00:13:03,310 Just an explanation of something we've 306 00:13:03,310 --> 00:13:06,610 had kind of across the bottom as an icon for a while. 307 00:13:06,610 --> 00:13:08,590 These things are called time value charts. 308 00:13:08,590 --> 00:13:11,530 They're useful if you want to make broad points 309 00:13:11,530 --> 00:13:13,960 about the use of time. 310 00:13:13,960 --> 00:13:17,210 This is a typical example, in a lot of ways. 311 00:13:17,210 --> 00:13:19,870 Here's a process that takes a while. 312 00:13:19,870 --> 00:13:22,870 Apparently there are two spans of four days 313 00:13:22,870 --> 00:13:27,250 of solid work in this process, as well as a lot of wait time 314 00:13:27,250 --> 00:13:30,625 and a lot of non-value added working time. 315 00:13:30,625 --> 00:13:32,500 One of the key things that this kind of chart 316 00:13:32,500 --> 00:13:34,743 will tell you about almost any process-- 317 00:13:34,743 --> 00:13:36,910 it's very bizarre if you come up with a process that 318 00:13:36,910 --> 00:13:38,260 isn't this way-- 319 00:13:38,260 --> 00:13:43,990 is that the big time savings are in the wait times 320 00:13:43,990 --> 00:13:46,030 and the non-value added times. 321 00:13:46,030 --> 00:13:51,700 And people were talking about the issue of standardizing work 322 00:13:51,700 --> 00:13:54,790 and eliminating creativity. 323 00:13:54,790 --> 00:14:00,610 What I always tell the engineers is this is you doing your work. 324 00:14:00,610 --> 00:14:03,130 We almost don't care how you do it because it's 325 00:14:03,130 --> 00:14:05,890 a tiny percentage of the time. 326 00:14:05,890 --> 00:14:08,140 To keep the customers happy, we got to take this out 327 00:14:08,140 --> 00:14:10,640 and we've got to take this out , and that's what we're going 328 00:14:10,640 --> 00:14:12,860 to worry about. 329 00:14:12,860 --> 00:14:16,480 And in a way, the same goes with medical care. 330 00:14:16,480 --> 00:14:20,800 A lot of the customer dissatisfaction 331 00:14:20,800 --> 00:14:23,940 comes from this kind of stuff. 332 00:14:23,940 --> 00:14:28,500 If my son gets injured, I take him to the hospital. 333 00:14:28,500 --> 00:14:32,430 I was absolutely delighted with the 45 seconds of care 334 00:14:32,430 --> 00:14:34,590 that he received in our nine-hour visit. 335 00:14:34,590 --> 00:14:38,740 So I was not going to tell the old male nurse who 336 00:14:38,740 --> 00:14:41,310 was really good at stitches how to do his job. 337 00:14:41,310 --> 00:14:44,070 I was delighted with his job. 338 00:14:44,070 --> 00:14:46,620 But there was a bunch of other processes involved 339 00:14:46,620 --> 00:14:48,270 in that particular experience that 340 00:14:48,270 --> 00:14:50,880 could have used improvement, which gets 341 00:14:50,880 --> 00:14:52,980 to the issue of pulling value. 342 00:14:52,980 --> 00:14:54,660 I was the customer there. 343 00:14:54,660 --> 00:14:58,240 I didn't really perceive value pull in that experience, 344 00:14:58,240 --> 00:15:01,160 although I've had some where I did. 345 00:15:01,160 --> 00:15:02,587 So what does this mean? 346 00:15:02,587 --> 00:15:04,170 This is a kind of a difficult concept, 347 00:15:04,170 --> 00:15:06,908 and it exists on a couple of scales. 348 00:15:06,908 --> 00:15:08,700 Sometimes the easiest way to think about it 349 00:15:08,700 --> 00:15:10,110 is to think of its opposite. 350 00:15:10,110 --> 00:15:14,160 A push system is an old-fashioned mass production 351 00:15:14,160 --> 00:15:15,960 system where what you're trying to do 352 00:15:15,960 --> 00:15:20,160 is optimize the productivity of each step. 353 00:15:20,160 --> 00:15:22,680 What that means is we set up our machine, 354 00:15:22,680 --> 00:15:26,608 or we set up our process, and we push a lot of material 355 00:15:26,608 --> 00:15:28,650 through it because that is the most efficient way 356 00:15:28,650 --> 00:15:31,710 to do most things is to set something up 357 00:15:31,710 --> 00:15:34,230 and then just bang stuff through it. 358 00:15:34,230 --> 00:15:36,300 And people sort of intuitively understand that. 359 00:15:36,300 --> 00:15:38,640 There was a question of why don't we just automate it? 360 00:15:38,640 --> 00:15:41,880 Well, to automate something, you have to set up a process, 361 00:15:41,880 --> 00:15:43,620 buy expensive equipment, and then 362 00:15:43,620 --> 00:15:46,830 when you're all done after doing all that prep work, voom, 363 00:15:46,830 --> 00:15:48,810 the shoes just fly out the end, and you 364 00:15:48,810 --> 00:15:50,460 can build huge piles of them. 365 00:15:50,460 --> 00:15:52,440 But then the question is, do people want them? 366 00:15:52,440 --> 00:15:53,970 Are they good? 367 00:15:53,970 --> 00:15:55,170 Is there quality issues? 368 00:15:55,170 --> 00:15:58,230 Is there issues if you want to change the style, 369 00:15:58,230 --> 00:16:01,320 how hard is that to do? 370 00:16:01,320 --> 00:16:02,670 So that's a push system. 371 00:16:02,670 --> 00:16:07,680 A pull system is kind of the conceptual opposite of that. 372 00:16:07,680 --> 00:16:10,440 Stuff only happens if the customers want it, 373 00:16:10,440 --> 00:16:12,750 and it happens, hopefully, quickly enough 374 00:16:12,750 --> 00:16:16,480 that the customer is satisfied by the process. 375 00:16:16,480 --> 00:16:19,270 So it's triggered by the customer. 376 00:16:19,270 --> 00:16:21,090 And that is what New Balance is really 377 00:16:21,090 --> 00:16:23,700 striving for they had a three-day turnaround. 378 00:16:23,700 --> 00:16:28,230 So the shoe store ordered shoes, and in three days 379 00:16:28,230 --> 00:16:31,350 the shoes were manufactured all the way from raw materials 380 00:16:31,350 --> 00:16:32,790 and sent to the shoe store. 381 00:16:32,790 --> 00:16:37,320 This both results in and necessitates a smooth flow 382 00:16:37,320 --> 00:16:41,670 with relatively modest batches and relatively small wait time, 383 00:16:41,670 --> 00:16:43,500 or it doesn't work. 384 00:16:43,500 --> 00:16:46,333 So as sort of an ideal, it's worth trying for. 385 00:16:46,333 --> 00:16:48,000 And inherently there's very little waste 386 00:16:48,000 --> 00:16:49,800 in that kind of system. 387 00:16:49,800 --> 00:16:53,100 Trouble is, of course, that's relatively hard to do. 388 00:16:53,100 --> 00:16:54,900 So let's think about what we need 389 00:16:54,900 --> 00:16:59,340 to do to do a pull system like we saw at New Balance. 390 00:16:59,340 --> 00:17:00,990 Pretty much requires flow. 391 00:17:00,990 --> 00:17:02,700 You don't have to have perfect flow, 392 00:17:02,700 --> 00:17:05,130 but if you don't have a system that flows reasonably well, 393 00:17:05,130 --> 00:17:07,200 it's very hard to make it a pull system. 394 00:17:07,200 --> 00:17:10,349 And you would like to have a fairly predictable cycle time. 395 00:17:10,349 --> 00:17:11,400 Here's a bunch of tools. 396 00:17:11,400 --> 00:17:13,319 We're going to get into those in a while. 397 00:17:13,319 --> 00:17:16,380 The point, really, with this slide is that to have pull, 398 00:17:16,380 --> 00:17:19,440 you need if not all of these, most of them. 399 00:17:19,440 --> 00:17:22,089 You have to have a pretty Lean system. 400 00:17:22,089 --> 00:17:24,480 And again, we'll get into what those are. 401 00:17:24,480 --> 00:17:27,780 And a pull system starts with the customer, and essentially 402 00:17:27,780 --> 00:17:30,780 the signals work backwards through the system. 403 00:17:30,780 --> 00:17:31,920 What does that mean? 404 00:17:31,920 --> 00:17:33,810 From the customer's point of view, 405 00:17:33,810 --> 00:17:36,840 it means that either the cycle time 406 00:17:36,840 --> 00:17:39,180 is lower than my expectation for when 407 00:17:39,180 --> 00:17:41,100 my product will be delivered or when 408 00:17:41,100 --> 00:17:42,600 my service will be delivered. 409 00:17:42,600 --> 00:17:45,360 Or there's inventory in the system. 410 00:17:45,360 --> 00:17:48,540 As a for example, I ordered a computer a while ago, 411 00:17:48,540 --> 00:17:49,980 that computer right there. 412 00:17:49,980 --> 00:17:53,133 And I sort of mentally budgeted a couple weeks. 413 00:17:53,133 --> 00:17:55,050 You get the new computer, put the files on it, 414 00:17:55,050 --> 00:17:57,420 get used to the system. 415 00:17:57,420 --> 00:17:59,730 I ordered it online. 416 00:17:59,730 --> 00:18:05,970 And it came in from China in less than 48 hours, 417 00:18:05,970 --> 00:18:07,230 less than two days. 418 00:18:07,230 --> 00:18:09,300 So to me, that was a pull system. 419 00:18:09,300 --> 00:18:12,420 It was way inside my expectations. 420 00:18:12,420 --> 00:18:18,240 So as far as it felt, I wanted it and it came. 421 00:18:18,240 --> 00:18:20,498 So that's a very nice system. 422 00:18:20,498 --> 00:18:22,290 The system was actually originally invented 423 00:18:22,290 --> 00:18:23,220 by Dell Computer. 424 00:18:23,220 --> 00:18:24,690 Apple's really good at it. 425 00:18:24,690 --> 00:18:27,570 Dell still is, actually, although interestingly, 426 00:18:27,570 --> 00:18:32,160 their market has shifted more to servers and industrial uses 427 00:18:32,160 --> 00:18:35,190 of computers than consumers. 428 00:18:35,190 --> 00:18:38,640 But they were one of the original inventors 429 00:18:38,640 --> 00:18:41,640 of essentially ordering something on the web, 430 00:18:41,640 --> 00:18:44,790 and then having it built for you in a process that's 431 00:18:44,790 --> 00:18:46,710 not really automated, but the process 432 00:18:46,710 --> 00:18:49,680 itself creates your computer. 433 00:18:49,680 --> 00:18:51,810 People don't think about the McManus order. 434 00:18:51,810 --> 00:18:53,670 It's just you order that computer. 435 00:18:53,670 --> 00:18:55,410 It goes into the system. 436 00:18:55,410 --> 00:18:59,560 That initiates the production of the computer, 437 00:18:59,560 --> 00:19:02,520 which goes very fast so it can ship the same day. 438 00:19:02,520 --> 00:19:05,140 And then they have a really nice supply chain operation 439 00:19:05,140 --> 00:19:07,950 so it can come all the way from China in about 24 hours, 440 00:19:07,950 --> 00:19:10,320 and you get your computer. 441 00:19:10,320 --> 00:19:14,320 This has become pretty standard for a lot of consumer products. 442 00:19:14,320 --> 00:19:15,940 So the bar has gone up. 443 00:19:15,940 --> 00:19:18,540 People are no longer mega impressed by this, 444 00:19:18,540 --> 00:19:20,310 but it's actually pretty impressive 445 00:19:20,310 --> 00:19:25,810 that people can do that on the customer end. 446 00:19:25,810 --> 00:19:28,930 A very useful concept when thinking about pull 447 00:19:28,930 --> 00:19:30,720 is takt time. 448 00:19:30,720 --> 00:19:33,930 If I want one, I'm really thinking about what's 449 00:19:33,930 --> 00:19:35,880 the cycle time I expect? 450 00:19:35,880 --> 00:19:38,080 I want a computer inside of a week, 451 00:19:38,080 --> 00:19:40,530 so if I get it in two days, I'm delighted. 452 00:19:40,530 --> 00:19:43,440 But if I'm Dell's current customers-- 453 00:19:43,440 --> 00:19:44,850 they're people like Google-- 454 00:19:44,850 --> 00:19:48,780 and I want 100,000 computers, then 455 00:19:48,780 --> 00:19:50,280 what we have to start thinking about 456 00:19:50,280 --> 00:19:52,350 is what is the pace at which I want 457 00:19:52,350 --> 00:19:54,268 those computers to show up? 458 00:19:54,268 --> 00:19:56,310 And that's a really simple calculation, actually, 459 00:19:56,310 --> 00:19:58,740 and one that's very much worth doing even 460 00:19:58,740 --> 00:20:02,400 in processes that are less predictable than production. 461 00:20:02,400 --> 00:20:03,900 Things like engineering and hospital 462 00:20:03,900 --> 00:20:06,370 processes, this is actually very useful to do, 463 00:20:06,370 --> 00:20:08,850 which is to just think about what's your available time-- 464 00:20:08,850 --> 00:20:11,010 a day, a week, a year, whatever it is-- 465 00:20:11,010 --> 00:20:15,250 and what is the customer demand during that time. 466 00:20:15,250 --> 00:20:22,680 So if I have a day and I have to see 16 patients, 467 00:20:22,680 --> 00:20:24,260 what's my takt time? 468 00:20:24,260 --> 00:20:28,620 Eight hour day, 16 patients, half hour. 469 00:20:28,620 --> 00:20:31,920 If I'm not seeing a patient every half hour, I'm behind. 470 00:20:31,920 --> 00:20:35,790 It's not rocket science, but it's 471 00:20:35,790 --> 00:20:38,970 a way of understanding what the customer demand is, 472 00:20:38,970 --> 00:20:42,540 and if I hope to have anything resembling a pull system, what 473 00:20:42,540 --> 00:20:45,150 kind of time constraints I have to stay within. 474 00:20:45,150 --> 00:20:48,160 So that's a very simple calculation here, 475 00:20:48,160 --> 00:20:50,250 which is basically the same one that we 476 00:20:50,250 --> 00:20:54,360 did for our medical example. 477 00:20:54,360 --> 00:20:59,870 This is for 40 orders across a year of 235 working days. 478 00:20:59,870 --> 00:21:00,940 Same idea. 479 00:21:00,940 --> 00:21:04,980 This is basically what is the pace 480 00:21:04,980 --> 00:21:10,430 at which we need to do work to satisfy the customer need? 481 00:21:10,430 --> 00:21:13,500 AUDIENCE: So that's presuming that this is like on average. 482 00:21:13,500 --> 00:21:16,220 But what if all 40 orders came in on the same day? 483 00:21:16,220 --> 00:21:16,940 AUDIENCE: Right. 484 00:21:16,940 --> 00:21:18,200 Yeah, that's right. 485 00:21:18,200 --> 00:21:20,930 Takt time is an average. 486 00:21:20,930 --> 00:21:25,640 The issue of irregularity we're going to explore a little later 487 00:21:25,640 --> 00:21:26,690 in the class. 488 00:21:26,690 --> 00:21:28,910 It definitely makes things harder. 489 00:21:28,910 --> 00:21:30,920 On the other hand, if you're dead on average, 490 00:21:30,920 --> 00:21:34,130 then no amount of variation is going to help you. 491 00:21:34,130 --> 00:21:35,870 Variation almost never helps you. 492 00:21:35,870 --> 00:21:37,850 So you start here. 493 00:21:37,850 --> 00:21:41,570 But yeah, there is that issue of what if-- 494 00:21:41,570 --> 00:21:44,503 sorry for the aside, but I was visiting Walmart, 495 00:21:44,503 --> 00:21:45,920 one of their distribution centers. 496 00:21:45,920 --> 00:21:47,360 They have an awesome supply chain. 497 00:21:47,360 --> 00:21:50,100 They're very, very good at what they do. 498 00:21:50,100 --> 00:21:53,550 They had this enormous warehouse in the middle of nowhere 499 00:21:53,550 --> 00:21:57,450 with about a cubic mile of lawn mowers in it. 500 00:21:57,450 --> 00:21:59,040 I exaggerate a little bit, but I think 501 00:21:59,040 --> 00:22:04,347 it was a good chunk of a mile long, only maybe 100 feet wide, 502 00:22:04,347 --> 00:22:06,930 and as many lawn mowers as you could stack before they started 503 00:22:06,930 --> 00:22:08,220 crushing each other. 504 00:22:08,220 --> 00:22:10,480 Not very Lean, right? 505 00:22:10,480 --> 00:22:12,670 But that's because everyone wants a lawn mower when? 506 00:22:15,450 --> 00:22:17,700 Yeah, like May, right? 507 00:22:17,700 --> 00:22:21,750 So sometimes you're just screwed. 508 00:22:21,750 --> 00:22:26,070 But on the other hand, the reason 509 00:22:26,070 --> 00:22:28,500 that they stack the lawn mowers up into the big cube 510 00:22:28,500 --> 00:22:30,780 is because the lawn mower factory was working to takt. 511 00:22:30,780 --> 00:22:32,430 They just figured out, over the whole year, 512 00:22:32,430 --> 00:22:33,680 how many lawn mowers they did. 513 00:22:33,680 --> 00:22:35,940 And they were just banging them out real efficiently 514 00:22:35,940 --> 00:22:37,860 at that rate, and they just decided 515 00:22:37,860 --> 00:22:40,080 that the best thing to do, although they 516 00:22:40,080 --> 00:22:42,690 weren't happy about it, was to pile them up and sell them 517 00:22:42,690 --> 00:22:43,470 all in May. 518 00:22:43,470 --> 00:22:48,870 So the takt still, actually, over a year, made sense. 519 00:22:48,870 --> 00:22:52,090 Just happened that they had to keep them all until May. 520 00:22:52,090 --> 00:22:52,590 OK. 521 00:22:52,590 --> 00:22:56,780 So what's our takt time at our hot dog stand? 522 00:22:56,780 --> 00:22:59,717 If we had 50 customers, what would our takt time be? 523 00:22:59,717 --> 00:23:02,050 Somebody that's got a calculator they can bang that out? 524 00:23:02,050 --> 00:23:06,010 Four hours, which is some number of minutes, which 525 00:23:06,010 --> 00:23:07,630 is some number of seconds. 526 00:23:07,630 --> 00:23:09,470 12.5 customers per hour. 527 00:23:09,470 --> 00:23:12,460 So how many hours per customer? 528 00:23:12,460 --> 00:23:15,290 And let's try to express that in seconds. 529 00:23:15,290 --> 00:23:15,790 Let's see. 530 00:23:15,790 --> 00:23:18,790 We got 240 minutes. 531 00:23:18,790 --> 00:23:20,620 So we got about four and 1/2 minutes. 532 00:23:20,620 --> 00:23:21,880 How many seconds is that? 533 00:23:21,880 --> 00:23:24,100 Can somebody multiply that by 60? 534 00:23:24,100 --> 00:23:25,505 280 seconds. 535 00:23:29,420 --> 00:23:30,500 All right. 536 00:23:30,500 --> 00:23:35,050 And if we wanted 75 customers, we'd need even less 192. 537 00:23:35,050 --> 00:23:35,810 OK. 538 00:23:35,810 --> 00:23:37,850 That's for 50. 539 00:23:37,850 --> 00:23:39,290 192. 540 00:23:39,290 --> 00:23:47,700 475 So we've computed a takt time and a cycle time 541 00:23:47,700 --> 00:23:49,440 for our hot dog stand. 542 00:23:49,440 --> 00:23:51,090 And first order doesn't look good. 543 00:23:51,090 --> 00:23:53,850 Our cycle time is longer than our takt time. 544 00:23:53,850 --> 00:23:57,510 But that's a little deceptive because there's two workers. 545 00:23:57,510 --> 00:24:00,870 It takes one hot dog, this one to order of two hot dogs, 546 00:24:00,870 --> 00:24:03,750 about this many seconds to get through the system. 547 00:24:03,750 --> 00:24:06,460 But it's trading off between two workers. 548 00:24:06,460 --> 00:24:08,820 So we can't just directly compare those. 549 00:24:08,820 --> 00:24:11,220 Still, it probably makes us a little nervous 550 00:24:11,220 --> 00:24:16,890 that our takt time, the sort of average time we have to hit, 551 00:24:16,890 --> 00:24:20,250 is considerably lower than the amount of time 552 00:24:20,250 --> 00:24:22,870 it actually takes. 553 00:24:22,870 --> 00:24:26,510 We'll save that thought for a little later 554 00:24:26,510 --> 00:24:28,820 and move on to the thought of balanced work. 555 00:24:28,820 --> 00:24:31,980 Actually, again, in this example it's pretty easy. 556 00:24:31,980 --> 00:24:32,900 We've got two workers. 557 00:24:32,900 --> 00:24:34,520 If their work is perfectly balanced, 558 00:24:34,520 --> 00:24:38,150 they're kind of doing half the work each. 559 00:24:38,150 --> 00:24:39,470 Is that the case? 560 00:24:39,470 --> 00:24:40,560 Don't know. 561 00:24:40,560 --> 00:24:41,060 OK. 562 00:24:41,060 --> 00:24:44,900 So we have this example of a product that needs 563 00:24:44,900 --> 00:24:48,320 to be delivered every six days. 564 00:24:48,320 --> 00:24:52,017 If it takes 30 days to build, how can we do it? 565 00:24:52,017 --> 00:24:53,850 Well, one thing we could do is split the job 566 00:24:53,850 --> 00:24:56,910 the job up into five pieces. 567 00:24:56,910 --> 00:24:59,940 And then they should take six days each. 568 00:24:59,940 --> 00:25:01,500 So that would work. 569 00:25:01,500 --> 00:25:06,240 And as unit one goes into the production there, 570 00:25:06,240 --> 00:25:08,850 it takes a couple of days to get things started. 571 00:25:08,850 --> 00:25:11,143 Well, [INAUDIBLE] has taken off by itself. 572 00:25:11,143 --> 00:25:12,810 I don't have to push the button anymore. 573 00:25:12,810 --> 00:25:17,010 Every six days a unit comes out, so we're making our takt time, 574 00:25:17,010 --> 00:25:18,930 even though our cycle time is longer. 575 00:25:18,930 --> 00:25:23,940 But we have five steps, five workers, five units of work. 576 00:25:23,940 --> 00:25:25,260 And they're balanced. 577 00:25:25,260 --> 00:25:26,130 They take the same. 578 00:25:26,130 --> 00:25:31,920 If one of those processes took eight days, would that work? 579 00:25:31,920 --> 00:25:32,640 No. 580 00:25:32,640 --> 00:25:34,020 It would take eight days. 581 00:25:34,020 --> 00:25:37,800 get The takt time would basically go to eight days. 582 00:25:37,800 --> 00:25:40,660 So again, not rocket science, fairly simple. 583 00:25:40,660 --> 00:25:43,057 But if we look at existing processes, 584 00:25:43,057 --> 00:25:44,640 if we're plotting this out, it's easy. 585 00:25:44,640 --> 00:25:46,020 If we look at existing processes, 586 00:25:46,020 --> 00:25:47,820 they're very rarely balanced. 587 00:25:47,820 --> 00:25:51,330 There usually is a bottleneck process 588 00:25:51,330 --> 00:25:54,420 or a problem where one step takes longer 589 00:25:54,420 --> 00:25:56,020 than some of the other ones. 590 00:25:56,020 --> 00:25:58,410 So balance is a key concept. 591 00:25:58,410 --> 00:25:59,110 Standard work. 592 00:25:59,110 --> 00:26:01,230 We've talked quite a lot about the fact 593 00:26:01,230 --> 00:26:07,200 that if we can't standardize work, then 594 00:26:07,200 --> 00:26:08,970 it's very difficult to improve it. 595 00:26:08,970 --> 00:26:12,840 Single piece flow is a key concept that's kind of integral 596 00:26:12,840 --> 00:26:15,990 to the idea of pull, that we're processing one unit at a time, 597 00:26:15,990 --> 00:26:18,430 just like that illustration, through all 598 00:26:18,430 --> 00:26:19,740 the steps to completion. 599 00:26:19,740 --> 00:26:21,240 That's kind of an ideal. 600 00:26:21,240 --> 00:26:23,880 It works great on a PowerPoint. 601 00:26:23,880 --> 00:26:25,860 Things are going one step at a time. 602 00:26:25,860 --> 00:26:26,820 No inventory. 603 00:26:26,820 --> 00:26:27,465 That's great. 604 00:26:27,465 --> 00:26:30,630 In practice it's sort of an ideal that we shoot towards. 605 00:26:30,630 --> 00:26:33,210 We try to keep our batches small. 606 00:26:33,210 --> 00:26:36,210 We try to make processes, when we can, 607 00:26:36,210 --> 00:26:39,660 line up so that the flow is single piece. 608 00:26:39,660 --> 00:26:42,120 That's kind of the opposite, again, of a batch and queue, 609 00:26:42,120 --> 00:26:43,860 where we make big batches of stuff 610 00:26:43,860 --> 00:26:48,980 and then it waits until the next step is ready to take it. 611 00:26:48,980 --> 00:26:53,020 This is a key conceptual tool for any kind of pull process. 612 00:26:53,020 --> 00:26:55,060 It's called a manufacturing cell. 613 00:26:55,060 --> 00:26:59,350 And we actually saw these in action at New Balance. 614 00:26:59,350 --> 00:27:02,050 The classic cell is actually U-shaped. 615 00:27:02,050 --> 00:27:04,270 People actually arrange the stuff 616 00:27:04,270 --> 00:27:08,130 in a U on the factory floor because then the in and out 617 00:27:08,130 --> 00:27:09,850 are basically in the same place. 618 00:27:09,850 --> 00:27:11,560 Makes moving stuff around easier. 619 00:27:11,560 --> 00:27:15,400 There is even interesting rules about whether they should 620 00:27:15,400 --> 00:27:18,220 go clockwise or counterclockwise depending on whether you're 621 00:27:18,220 --> 00:27:20,152 right-handed or left-handed. 622 00:27:20,152 --> 00:27:21,860 If you want to get into the ergonometrics 623 00:27:21,860 --> 00:27:22,860 you can get very clever. 624 00:27:22,860 --> 00:27:26,380 But the basic idea is, OK, if we have some tasks that 625 00:27:26,380 --> 00:27:28,720 can be balanced, that can be made 626 00:27:28,720 --> 00:27:31,150 to take about the same time and that 627 00:27:31,150 --> 00:27:34,570 aren't dependent on a big batch machine like those cutters, 628 00:27:34,570 --> 00:27:37,570 then we can line them up, balance them 629 00:27:37,570 --> 00:27:39,460 so they take about the same time, 630 00:27:39,460 --> 00:27:42,970 and then put in a very interesting rule. 631 00:27:42,970 --> 00:27:45,610 Instead of a big, complicated scheduling system, 632 00:27:45,610 --> 00:27:49,210 we're going to say don't work unless the downstream process 633 00:27:49,210 --> 00:27:50,410 needs you to. 634 00:27:50,410 --> 00:27:52,990 And you can tell whether it needs you to-- 635 00:27:52,990 --> 00:27:54,400 you don't even have to ask-- 636 00:27:54,400 --> 00:27:56,620 by the fact that their inventory bin is empty. 637 00:27:56,620 --> 00:27:58,360 So very simple rule. 638 00:27:58,360 --> 00:28:00,970 So in practice, OK, everybody has inventory, 639 00:28:00,970 --> 00:28:06,370 nobody works until the next process 640 00:28:06,370 --> 00:28:08,560 in line, the internal or external customer, 641 00:28:08,560 --> 00:28:11,590 takes product away from here. 642 00:28:11,590 --> 00:28:13,690 Now task four starts to work. 643 00:28:13,690 --> 00:28:16,935 To do that, they start using their inventory, which 644 00:28:16,935 --> 00:28:18,310 means three starts working, which 645 00:28:18,310 --> 00:28:21,110 means two starts working, which means one starts working. 646 00:28:21,110 --> 00:28:23,120 So now everybody's working. 647 00:28:23,120 --> 00:28:24,900 Oh, and of course, now we're out here. 648 00:28:24,900 --> 00:28:28,940 So we have to go ask our internal or external supplier 649 00:28:28,940 --> 00:28:30,890 for more parts, for more stuff. 650 00:28:30,890 --> 00:28:33,260 If everybody takes exactly the same amount of time, 651 00:28:33,260 --> 00:28:35,710 my next click would get us back to the beginning. 652 00:28:35,710 --> 00:28:38,300 Everything would go back to everybody 653 00:28:38,300 --> 00:28:40,460 having work in front of them. 654 00:28:40,460 --> 00:28:42,530 So they stop working. 655 00:28:42,530 --> 00:28:44,450 No inventory. 656 00:28:44,450 --> 00:28:46,070 Direct customer pull. 657 00:28:46,070 --> 00:28:49,880 It's just triggered by the customer taking the item out 658 00:28:49,880 --> 00:28:52,250 of the inventory of one at the end. 659 00:28:52,250 --> 00:28:55,610 In the real world, some tasks, maybe the balance 660 00:28:55,610 --> 00:28:59,210 isn't perfect, these tasks finish first. 661 00:28:59,210 --> 00:29:01,640 They notice now that there is inventory sitting there 662 00:29:01,640 --> 00:29:04,077 because this task is not ready to take it, 663 00:29:04,077 --> 00:29:04,910 so they're finished. 664 00:29:04,910 --> 00:29:07,530 They stop working. 665 00:29:07,530 --> 00:29:09,527 That seems inefficient, but what good 666 00:29:09,527 --> 00:29:10,860 does it do if they keep working? 667 00:29:10,860 --> 00:29:13,105 What happens if they keep working? 668 00:29:13,105 --> 00:29:13,980 AUDIENCE: [INAUDIBLE] 669 00:29:13,980 --> 00:29:15,980 PROFESSOR: Yeah, it just creates more inventory. 670 00:29:15,980 --> 00:29:18,100 We can just pile up inventory here. 671 00:29:18,100 --> 00:29:20,670 That would be a waste of overproduction. 672 00:29:20,670 --> 00:29:23,820 So that's very simple, but this is 673 00:29:23,820 --> 00:29:25,930 very practical in the manufacturing system. 674 00:29:25,930 --> 00:29:27,930 And conceptually, you can use it for things 675 00:29:27,930 --> 00:29:29,670 like engineering work. 676 00:29:29,670 --> 00:29:34,170 Rapid drawing release, people have set up cells. 677 00:29:34,170 --> 00:29:36,600 That horrible value stream we saw at the very beginning 678 00:29:36,600 --> 00:29:38,100 with all the lines going everywhere, 679 00:29:38,100 --> 00:29:40,260 well, that was to get one drawing out. 680 00:29:40,260 --> 00:29:42,390 So if you just line the engineers up like this, 681 00:29:42,390 --> 00:29:44,560 makes as much sense ever as anything else. 682 00:29:44,560 --> 00:29:45,960 It may not be perfectly efficient 683 00:29:45,960 --> 00:29:48,000 from the engineer's point of view, 684 00:29:48,000 --> 00:29:51,990 but it minimizes, without any complicated control system, 685 00:29:51,990 --> 00:29:54,550 the time the product spends in the system. 686 00:29:54,550 --> 00:29:57,160 And that's the advantage of a pull system. 687 00:29:57,160 --> 00:30:01,180 Another example of pull as a kanban system. 688 00:30:01,180 --> 00:30:03,520 We talked about that a little bit before. 689 00:30:03,520 --> 00:30:06,760 Essentially, same rule. 690 00:30:06,760 --> 00:30:10,060 Appearance of a kanban bin means that the downstream process 691 00:30:10,060 --> 00:30:12,790 wants parts. 692 00:30:12,790 --> 00:30:15,970 Kanban card or bin, some kind of signal. 693 00:30:15,970 --> 00:30:18,260 The difference here is that we're no longer in a cell. 694 00:30:18,260 --> 00:30:20,360 We're no longer right next to the other process. 695 00:30:20,360 --> 00:30:22,210 So we need some kind of signal. 696 00:30:22,210 --> 00:30:25,240 Now it could be a complicated order with legal and pages, 697 00:30:25,240 --> 00:30:28,570 or it could be a bin or a card, something simple 698 00:30:28,570 --> 00:30:31,540 that we've prearranged that says, I need work. 699 00:30:31,540 --> 00:30:33,970 I need you to do work for me. 700 00:30:33,970 --> 00:30:36,670 Of course, you have to have standard processes for this. 701 00:30:36,670 --> 00:30:39,760 You have to agree what this card means. 702 00:30:39,760 --> 00:30:42,760 But it does allow us to control even a fairly complicated 703 00:30:42,760 --> 00:30:46,330 system visually, and even if it's geographically 704 00:30:46,330 --> 00:30:48,190 or organizationally separated. 705 00:30:48,190 --> 00:30:49,790 Again, a very simple example. 706 00:30:49,790 --> 00:30:53,020 If we have a supplier and a manufacturer-- 707 00:30:53,020 --> 00:30:55,030 in this case, we have physical bins. 708 00:30:55,030 --> 00:30:57,250 In the New Balance video, you saw 709 00:30:57,250 --> 00:30:59,050 those plastic bins that everybody was 710 00:30:59,050 --> 00:31:00,940 using to move the parts around. 711 00:31:00,940 --> 00:31:05,260 It's very real, very direct and physical. 712 00:31:05,260 --> 00:31:08,020 If we run out, we send the empty bin to the supplier. 713 00:31:08,020 --> 00:31:09,350 They know what that means. 714 00:31:09,350 --> 00:31:11,227 It means fill it up. 715 00:31:11,227 --> 00:31:13,060 So they work on it and they fill up the bin. 716 00:31:13,060 --> 00:31:15,340 When it's full, shipping department 717 00:31:15,340 --> 00:31:18,130 knows that means to take it away. 718 00:31:18,130 --> 00:31:20,170 When it shows up, the manufacturer 719 00:31:20,170 --> 00:31:22,180 uses it until the bin is empty. 720 00:31:22,180 --> 00:31:23,350 Back to the supplier. 721 00:31:23,350 --> 00:31:25,180 Self-perpetuating system. 722 00:31:25,180 --> 00:31:28,180 Even if these systems are not particularly well-balanced, 723 00:31:28,180 --> 00:31:32,000 or even if demand is intermittent, 724 00:31:32,000 --> 00:31:37,450 it will operate with no outside interference, no management 725 00:31:37,450 --> 00:31:39,910 effort, and minimum inventory. 726 00:31:39,910 --> 00:31:43,360 About the only trick in this is essentially how many bins. 727 00:31:43,360 --> 00:31:46,420 If you only have one bin, what's the problem? 728 00:31:46,420 --> 00:31:47,530 Yeah. 729 00:31:47,530 --> 00:31:50,445 You have to wait, right? 730 00:31:50,445 --> 00:31:51,820 You have to wait for the supplier 731 00:31:51,820 --> 00:31:53,440 to be finished before you can start working 732 00:31:53,440 --> 00:31:55,390 again, which might be OK if it's a low demand 733 00:31:55,390 --> 00:31:57,640 product or something. 734 00:31:57,640 --> 00:32:00,190 In this example, we have four, so it kind of continues 735 00:32:00,190 --> 00:32:01,012 round and round. 736 00:32:01,012 --> 00:32:02,470 Worst thing that can happen here is 737 00:32:02,470 --> 00:32:06,220 if there's no demand, four bins worth pile up 738 00:32:06,220 --> 00:32:09,490 because the supplier will keep moving until all the are full. 739 00:32:09,490 --> 00:32:10,982 So we could pile up four bins. 740 00:32:10,982 --> 00:32:12,940 That's probably not horrible unless, of course, 741 00:32:12,940 --> 00:32:14,590 they're really expensive things. 742 00:32:14,590 --> 00:32:17,440 So there's essentially a design knob 743 00:32:17,440 --> 00:32:20,150 on the system, which is how many bins, but that's it. 744 00:32:20,150 --> 00:32:21,970 It's very simple. 745 00:32:21,970 --> 00:32:26,980 And we can tweak it as necessary by changing the number of bins. 746 00:32:26,980 --> 00:32:27,520 OK. 747 00:32:27,520 --> 00:32:28,780 So that's kanban. 748 00:32:28,780 --> 00:32:31,000 Kanban itself is kind of a visual control. 749 00:32:31,000 --> 00:32:33,650 You see an empty bin, you work. 750 00:32:33,650 --> 00:32:38,410 There's other kinds of visual controls 751 00:32:38,410 --> 00:32:40,900 that are referred to generically, again, 752 00:32:40,900 --> 00:32:43,510 with another Japanese word, the word Andon. 753 00:32:43,510 --> 00:32:46,150 Andon is kind of an adjective that means 754 00:32:46,150 --> 00:32:48,320 visual control, more or less. 755 00:32:48,320 --> 00:32:51,100 And you can have, for example, Andon lights. 756 00:32:51,100 --> 00:32:55,060 Again, in the video, you may have noticed in the background 757 00:32:55,060 --> 00:32:57,610 some lights like this. 758 00:32:57,610 --> 00:33:00,700 Very standard practice now in manufacturing 759 00:33:00,700 --> 00:33:03,160 to have a little light tower like that. 760 00:33:03,160 --> 00:33:05,140 Red means the machine is broken or there's 761 00:33:05,140 --> 00:33:06,370 some serious problem. 762 00:33:06,370 --> 00:33:07,540 It needs attention. 763 00:33:07,540 --> 00:33:09,070 Green means it's going fine. 764 00:33:09,070 --> 00:33:11,410 Yellow means maybe were behind a little bit. 765 00:33:11,410 --> 00:33:14,320 It allows you to just look across the factory floor, 766 00:33:14,320 --> 00:33:14,980 see the state. 767 00:33:14,980 --> 00:33:16,690 If everything's green, it's dandy. 768 00:33:16,690 --> 00:33:18,880 If there's red stuff, you got to go fix it. 769 00:33:18,880 --> 00:33:21,790 If there's yellows, that tells you what to worry about. 770 00:33:21,790 --> 00:33:24,940 It's only really useful if you do something about it, though. 771 00:33:24,940 --> 00:33:28,590 It's part of an active management system. 772 00:33:28,590 --> 00:33:30,180 Another one is the Andon cord. 773 00:33:30,180 --> 00:33:33,340 This is the famous counterintuitive Japanese 774 00:33:33,340 --> 00:33:36,940 practice of basically having a cord in front of every employee 775 00:33:36,940 --> 00:33:40,630 that they can pull that stops the entire factory. 776 00:33:40,630 --> 00:33:46,930 This is very scary, even to this day, to North American managers 777 00:33:46,930 --> 00:33:50,910 because any disgruntled person can pull the cord 778 00:33:50,910 --> 00:33:52,300 and stop the entire factory. 779 00:33:52,300 --> 00:33:53,820 What's that about? 780 00:33:53,820 --> 00:33:56,250 But of course, the way they use that is to draw attention 781 00:33:56,250 --> 00:33:57,900 to a serious problem. 782 00:33:57,900 --> 00:34:02,730 If a quality problem is happening, the parts don't fit, 783 00:34:02,730 --> 00:34:07,530 that's unacceptable at a high quality production operation. 784 00:34:07,530 --> 00:34:10,889 So they stop the line because if they don't, that defective part 785 00:34:10,889 --> 00:34:12,960 is going to continue down the line, 786 00:34:12,960 --> 00:34:16,020 get more parts piled on top of it until you have, 787 00:34:16,020 --> 00:34:17,760 essentially, a defective car. 788 00:34:17,760 --> 00:34:20,280 So a washer or whatever is turned into a big problem. 789 00:34:20,280 --> 00:34:22,230 It's actually better to stop and fix it. 790 00:34:22,230 --> 00:34:24,239 It also provides a lot of motivation. 791 00:34:24,239 --> 00:34:26,100 Everybody comes over and fixes it right away 792 00:34:26,100 --> 00:34:28,179 because the line is stopped. 793 00:34:28,179 --> 00:34:31,290 So the team leader, engineers, whoever 794 00:34:31,290 --> 00:34:34,710 needs to come on over and solve the problem. 795 00:34:38,139 --> 00:34:41,900 That's the Andon cord concept. 796 00:34:41,900 --> 00:34:43,860 Medical example. 797 00:34:43,860 --> 00:34:45,830 Virginia Mason Medical Center. 798 00:34:45,830 --> 00:34:50,000 They have essentially a policy that 799 00:34:50,000 --> 00:34:55,500 says that any of their 5,000 employees can-- 800 00:34:55,500 --> 00:34:57,380 they don't exactly stop seeing patients-- 801 00:34:57,380 --> 00:35:00,260 but all attention is paid to this person 802 00:35:00,260 --> 00:35:05,450 if they declare it an Andon cord action. 803 00:35:05,450 --> 00:35:07,280 If there is a patient safety issue that 804 00:35:07,280 --> 00:35:09,240 needs to be dealt with immediately, 805 00:35:09,240 --> 00:35:12,390 they can get as much attention as they need. 806 00:35:12,390 --> 00:35:15,870 And this happens quite a bit. 807 00:35:15,870 --> 00:35:20,200 But it has also surfaced a lot of problems. 808 00:35:20,200 --> 00:35:22,330 And the real key here, by the way, 809 00:35:22,330 --> 00:35:26,688 is not that this is a very good, quick problem-solving method, 810 00:35:26,688 --> 00:35:28,230 because it isn't-- it stops the line. 811 00:35:28,230 --> 00:35:29,813 It could be potentially very expensive 812 00:35:29,813 --> 00:35:31,360 over something very small-- 813 00:35:31,360 --> 00:35:33,390 but that it surface problems. 814 00:35:33,390 --> 00:35:35,590 If you have a problem of this magnitude, 815 00:35:35,590 --> 00:35:37,140 you don't want to see it again. 816 00:35:37,140 --> 00:35:40,530 So it surfaces the problem and allows 817 00:35:40,530 --> 00:35:43,380 people to pay attention to it, find the root cause, 818 00:35:43,380 --> 00:35:44,820 make it go away for good. 819 00:35:44,820 --> 00:35:45,390 OK. 820 00:35:45,390 --> 00:35:47,460 Final step, pursuing perfection. 821 00:35:47,460 --> 00:35:49,980 This is all about not quitting. 822 00:35:49,980 --> 00:35:51,720 There's always more ways to eliminate. 823 00:35:51,720 --> 00:35:54,660 There's always ways you can do things better. 824 00:35:54,660 --> 00:35:59,260 There's always ways you can get the employees more involved. 825 00:35:59,260 --> 00:36:01,780 So it's a continuous process. 826 00:36:01,780 --> 00:36:05,910 And we'll be talking about that more later. 827 00:36:05,910 --> 00:36:13,080 Another tool, a final tool for this afternoon, five whys. 828 00:36:13,080 --> 00:36:15,180 The eight wastes and the six S's. 829 00:36:15,180 --> 00:36:20,760 These are the five whys, not the letter, the word. 830 00:36:20,760 --> 00:36:24,180 And the point of the five whys is to keep asking why. 831 00:36:24,180 --> 00:36:26,430 The fact that there's a number is actually kind of bad 832 00:36:26,430 --> 00:36:28,680 because you may not want to quit at five, 833 00:36:28,680 --> 00:36:30,280 and three might be enough. 834 00:36:30,280 --> 00:36:32,640 But here's an example. 835 00:36:32,640 --> 00:36:36,990 Possibly apocryphal example about the Jefferson Monument. 836 00:36:36,990 --> 00:36:39,330 Why is the Jefferson Monument deteriorating? 837 00:36:39,330 --> 00:36:41,740 Because they wash it all the time. 838 00:36:41,740 --> 00:36:43,140 Why is it washed all the time? 839 00:36:43,140 --> 00:36:45,690 Because it has bird poop on it. 840 00:36:45,690 --> 00:36:46,840 Why do they do that? 841 00:36:46,840 --> 00:36:49,170 Well, because the birds are eating the spiders. 842 00:36:49,170 --> 00:36:50,770 Why are there spiders? 843 00:36:50,770 --> 00:36:52,470 Because they're eating gnats. 844 00:36:52,470 --> 00:36:53,700 Why are there gnats? 845 00:36:53,700 --> 00:36:56,220 Because the lights are on all the time. 846 00:36:56,220 --> 00:37:00,933 Of course, I had somebody at a class a couple of times ago 847 00:37:00,933 --> 00:37:02,850 that said, why are the lights on all the time? 848 00:37:02,850 --> 00:37:04,500 Because it's in a bad neighborhood. 849 00:37:09,060 --> 00:37:11,340 So maybe five is not the right number. 850 00:37:11,340 --> 00:37:12,510 Maybe less is OK. 851 00:37:12,510 --> 00:37:14,560 Maybe you need to go deeper. 852 00:37:14,560 --> 00:37:18,070 So the five fundamentals kind of work together. 853 00:37:18,070 --> 00:37:20,640 Pursuing perfection may send you all the way back 854 00:37:20,640 --> 00:37:21,630 to the beginning. 855 00:37:21,630 --> 00:37:23,880 This is kind of a strategic view. 856 00:37:23,880 --> 00:37:26,280 The tactics-- and we're going to be talking about this 857 00:37:26,280 --> 00:37:28,380 for the rest of the week so I won't dwell on it-- 858 00:37:28,380 --> 00:37:33,300 are, in this loop, the Plan, Do, Study, Act loop. 859 00:37:33,300 --> 00:37:36,930 So whenever you find something you need to fix, 860 00:37:36,930 --> 00:37:39,180 you don't just willy nilly fix it. 861 00:37:39,180 --> 00:37:41,070 You go through an ordered process, 862 00:37:41,070 --> 00:37:44,310 and it is through a series of these ordered processes 863 00:37:44,310 --> 00:37:48,600 that you actually improve your value stream, 864 00:37:48,600 --> 00:37:52,710 achieve your flow, push towards pull. 865 00:37:52,710 --> 00:37:54,932 So we've actually gone through a bunch of stuff. 866 00:37:54,932 --> 00:37:56,640 Remember that huge chart at the beginning 867 00:37:56,640 --> 00:38:00,270 that nobody knew hardly any of the turns on? 868 00:38:00,270 --> 00:38:02,220 Well, now you have at least been introduced 869 00:38:02,220 --> 00:38:03,510 to all of these, quite a few. 870 00:38:06,747 --> 00:38:08,830 I don't know whether you've absorbed all of those. 871 00:38:08,830 --> 00:38:10,455 Hopefully at the top level, the concept 872 00:38:10,455 --> 00:38:15,060 of process, the concept of thinking about customer value 873 00:38:15,060 --> 00:38:19,340 has sunk in, and that some of the thoughts about making 874 00:38:19,340 --> 00:38:22,400 value flow, a couple of the simple tools 875 00:38:22,400 --> 00:38:24,440 have stuck, particularly-- not on the chart, 876 00:38:24,440 --> 00:38:28,490 but always a favorite of mine-- is that you've at least started 877 00:38:28,490 --> 00:38:30,530 to appreciate the complexity of what 878 00:38:30,530 --> 00:38:32,720 time means in these processes. 879 00:38:32,720 --> 00:38:36,050 And again, we'll be doing a considerable exercise on that 880 00:38:36,050 --> 00:38:37,500 starting in just a few minutes. 881 00:38:37,500 --> 00:38:40,270 So that's it for now.