1 00:00:00,240 --> 00:00:02,360 Hi, and welcome to the Neil Ashton Podcast. 2 00:00:03,000 --> 00:00:05,880 In each episode, we explain some of the fascinating ways 3 00:00:05,880 --> 00:00:09,080 that science and engineering are changing the world around us. 4 00:00:09,760 --> 00:00:12,720 We talk to leading engineers from elite level sports like 5 00:00:12,800 --> 00:00:16,840 cycling and Formula One to some of the world's top academics to 6 00:00:16,840 --> 00:00:20,445 understand how fluid dynamics, machine learning and 7 00:00:20,445 --> 00:00:22,960 supercomputing are bringing in a new era of discovery. 8 00:00:23,920 --> 00:00:27,040 We also hear some of their life stories, their career advice, 9 00:00:27,640 --> 00:00:30,040 the lessons they've learned on the way that I hope will be 10 00:00:30,040 --> 00:00:33,800 helpful to you too. So sit back and enjoy this 11 00:00:33,800 --> 00:00:40,840 episode. Hi, and welcome back to the Neil 12 00:00:40,840 --> 00:00:43,960 Ashton Podcast. I just finished recording 13 00:00:43,960 --> 00:00:48,240 today's episode with Professor Brian Launder, the incredible 14 00:00:48,240 --> 00:00:53,600 person who has truly been one of The Pioneers and legends of 15 00:00:53,600 --> 00:00:57,200 computational fluid dynamics and fluid dynamics more broadly. 16 00:00:57,280 --> 00:01:03,760 He is most well known for the k-epsilon model, for the LRR 17 00:01:03,920 --> 00:01:07,000 Reynolds-stress model. And I would argue, and I didn't 18 00:01:07,000 --> 00:01:10,200 fully appreciate it until after actually speaking to him, just 19 00:01:10,200 --> 00:01:16,200 the number of people that he supervised, mentored and 20 00:01:16,200 --> 00:01:19,200 brought, you know, these people like Bill Jones and Kemal 21 00:01:19,200 --> 00:01:23,176 Hanjalić and Michael Leschziner and Tim Craft and Hector 22 00:01:23,176 --> 00:01:26,840 Iacovides, who are people that themselves have gone on to 23 00:01:26,848 --> 00:01:29,055 supervise people. You know, this is the thing with 24 00:01:29,055 --> 00:01:31,238 academia. One of the core roles I believe 25 00:01:31,238 --> 00:01:35,080 of of as a professor is research, of course, but the 26 00:01:35,088 --> 00:01:39,490 people that they nurture and then bring on to go into 27 00:01:39,498 --> 00:01:42,050 industry, go into academia is the real credit. 28 00:01:42,058 --> 00:01:45,067 And so many people have benefited from from what he's 29 00:01:45,075 --> 00:01:46,754 done. It's hard to believe that most 30 00:01:46,762 --> 00:01:50,028 of this stuff was done in the 1970s, fifty years ago. 31 00:01:50,036 --> 00:01:53,770 Today we are still using the K epsilon model. 32 00:01:53,778 --> 00:01:57,175 If you're doing a CFD simulation of many different objects, 33 00:01:57,183 --> 00:02:00,632 whether you know heat transfer cars, engines, you're probably 34 00:02:00,640 --> 00:02:04,970 using what the k-epsilon or some variant of it or, or Reynolds 35 00:02:04,978 --> 00:02:07,332 stress model. And the fact that that was done 36 00:02:07,332 --> 00:02:10,062 in the 70s when computers, you know, he was telling us that you 37 00:02:10,062 --> 00:02:11,240 have to walk up to the building. 38 00:02:11,240 --> 00:02:16,240 There was no SSH remote access punch cards, you know, the codes 39 00:02:16,240 --> 00:02:18,600 were just being written. It was incredible. 40 00:02:19,320 --> 00:02:24,040 I genuinely in awe of the development that happened back 41 00:02:24,040 --> 00:02:28,840 then and it we we went through essentially his life story from 42 00:02:28,840 --> 00:02:32,360 his undergraduate years moving across to the US from Imperial 43 00:02:32,360 --> 00:02:36,440 to then MIT coming back to Imperial, the role of Brian 44 00:02:36,440 --> 00:02:39,800 Spalding, a key interesting one that we reflect on at the 45 00:02:39,800 --> 00:02:42,840 beginning and the end, you know, incredible at that time, the 46 00:02:42,840 --> 00:02:45,720 turbulent time of how the CHAM code came about and some of the 47 00:02:45,720 --> 00:02:50,584 tensions around that. His ultimate move to UMIST now 48 00:02:50,584 --> 00:02:54,917 the University of Manchester. And as I said, the the key 49 00:02:54,925 --> 00:02:58,060 developments around the eddy viscosity model, second moment 50 00:02:58,068 --> 00:03:01,082 closure models, but also things like the wall functions that I 51 00:03:01,090 --> 00:03:04,948 mentioned to him is now even more relevant in the context of 52 00:03:04,948 --> 00:03:07,947 wall-modelled LES etcetera. And just yeah, an incredible 53 00:03:07,955 --> 00:03:10,121 person. And I think some of the advice 54 00:03:10,129 --> 00:03:13,000 and lessons you get from it, it reminds you even the greatest 55 00:03:13,008 --> 00:03:15,480 people have some of the challenges that that we have. 56 00:03:15,488 --> 00:03:18,900 And, and they don't realize at the time the amazing work that 57 00:03:18,908 --> 00:03:23,320 they're doing. But yeah, really, really special 58 00:03:23,320 --> 00:03:25,624 person. And someone I really liked 59 00:03:25,632 --> 00:03:28,730 speaking to, you know, he's a fellow with the Royal Society. 60 00:03:28,738 --> 00:03:32,678 He's he's done so many things. He's you know, how many good 61 00:03:32,686 --> 00:03:35,752 citations, you know, 10s of thousands, super high h-index, 62 00:03:35,752 --> 00:03:39,190 all of all of the things, of course, that a top professor 63 00:03:39,198 --> 00:03:41,710 would have. So if you're looking for the 64 00:03:41,718 --> 00:03:45,235 pinnacle of people in the simulation world, quite a few of 65 00:03:45,235 --> 00:03:48,355 his people he supervised and worked with have now gone off 66 00:03:48,363 --> 00:03:50,776 into some of these major multibillion dollar companies. 67 00:03:50,784 --> 00:03:54,932 In fact, if you use these big simulation codes, many of them 68 00:03:54,940 --> 00:03:58,695 are the foundations are on the work that the professor Brian 69 00:03:58,703 --> 00:04:01,527 Launder was part of or, or or leading. 70 00:04:01,535 --> 00:04:04,745 And so, yeah, really interesting story as as we've ever. 71 00:04:04,753 --> 00:04:07,071 There was lots of things that I could have asked him. 72 00:04:07,079 --> 00:04:10,440 We did talk a little bit about some of the rivalry with 73 00:04:10,448 --> 00:04:13,644 Spalart, Menter and Wilcox and, of course, you know, Professor 74 00:04:13,644 --> 00:04:16,182 Launder is a gentleman. And then I genuinely think his 75 00:04:16,190 --> 00:04:19,094 focus was more on the research. But it's always interesting to 76 00:04:19,101 --> 00:04:22,375 us who used turbulence models to understand, you know, what was 77 00:04:22,383 --> 00:04:25,620 the difference between these these key figures, I guess at 78 00:04:25,628 --> 00:04:28,585 the time. But so I hope you enjoy this 79 00:04:28,593 --> 00:04:30,200 episode. I genuinely really did. 80 00:04:30,208 --> 00:04:33,116 I, I mentioned that I studied the University of Manchester. 81 00:04:33,124 --> 00:04:34,500 I was taught by Professor Launder's. 82 00:04:34,508 --> 00:04:37,240 So it was one of my good friends, Alastair West, who was 83 00:04:37,240 --> 00:04:41,154 also his last PhD student. And yeah, it's it was an honor 84 00:04:41,162 --> 00:04:44,460 to speak to him. And I hope you enjoyed this as 85 00:04:44,468 --> 00:04:48,456 much as I did speaking to him. So sit back and enjoy this 86 00:04:48,464 --> 00:04:50,029 episode with Professor Brian Launder. 87 00:04:50,037 --> 00:04:52,510 Thank you very much for agreeing to do this. 88 00:04:52,518 --> 00:04:55,482 I really appreciate you taking the time to speak to me today. 89 00:04:55,490 --> 00:04:59,220 And I thought maybe it would be good for the people listening 90 00:04:59,228 --> 00:05:03,002 to, you know, hear about your early days, where it all began 91 00:05:03,010 --> 00:05:06,136 and you starting off, you know, what did you do at university? 92 00:05:06,144 --> 00:05:07,595 What was your undergraduate studies? 93 00:05:07,603 --> 00:05:12,032 OK, well that doesn't sound a bad place to start. 94 00:05:12,040 --> 00:05:14,725 Let's let's begin with early 1961. 95 00:05:14,733 --> 00:05:20,288 I was in the final year of my bachelor's program at Imperial 96 00:05:20,296 --> 00:05:23,368 College. I found a topic that really 97 00:05:23,376 --> 00:05:28,310 interested me, convective heat transfer, and I was doing a 98 00:05:28,318 --> 00:05:32,254 final year project on flows related to boiling. 99 00:05:32,262 --> 00:05:36,280 When I heard that the new professor of heat transfer was 100 00:05:36,288 --> 00:05:40,248 going to talk about postgraduate opportunities, I was quite 101 00:05:40,256 --> 00:05:43,363 excited. I made sure to attend because I 102 00:05:43,363 --> 00:05:47,664 knew by then that I would like to do postgraduate work. 103 00:05:47,672 --> 00:05:53,535 So I went along to the meeting that he was to address, and he 104 00:05:53,543 --> 00:05:57,429 probably did talk about the range of opportunities that 105 00:05:57,437 --> 00:06:02,260 there might be. But what I remembered were a 106 00:06:02,268 --> 00:06:06,520 particular sentence or two that he presented us with. 107 00:06:06,528 --> 00:06:12,360 He said, you well know that the college is being rebuilt, and 108 00:06:12,368 --> 00:06:16,494 they're rebuilding it on the same site as the present 109 00:06:16,502 --> 00:06:20,440 building. This will inevitably mean 110 00:06:20,448 --> 00:06:25,819 interruptions and delays. So if you're looking for a fast 111 00:06:25,819 --> 00:06:31,105 PhD, you ought to go elsewhere. This was a very deflating 112 00:06:31,113 --> 00:06:35,872 statement from him. A day or two later, I asked my 113 00:06:35,880 --> 00:06:40,020 closest friend in the course guy called Hugh Kendrick what he 114 00:06:40,028 --> 00:06:44,318 planned to do. He chuckled, looked at me and 115 00:06:44,326 --> 00:06:46,738 said I'll probably go to America. 116 00:06:46,746 --> 00:06:51,376 I fancy doing a master's program at Yale or Princeton. 117 00:06:51,384 --> 00:06:55,744 Which time I interrupted him and said, you're talking rubbish. 118 00:06:55,752 --> 00:07:00,006 What would a top American university do, having a thicko 119 00:07:00,014 --> 00:07:02,722 like you? We both chuckled at that point 120 00:07:02,730 --> 00:07:07,290 and moved on to other things. But you know, when I was alone 121 00:07:07,298 --> 00:07:10,980 reflecting on this in the evening, I thought that's, 122 00:07:10,988 --> 00:07:17,800 that's a pretty neat idea. So I wrote to a half dozen 123 00:07:17,808 --> 00:07:23,076 universities in the USA seeking graduate or postgraduate 124 00:07:23,084 --> 00:07:26,344 engagement there, together with funding. 125 00:07:26,352 --> 00:07:30,525 Nothing happened for several weeks, of course, but then 126 00:07:30,533 --> 00:07:35,325 responses started to come in and I was absolutely delighted. 127 00:07:35,333 --> 00:07:40,080 Princeton offered me a name scholarship that would cover my 128 00:07:40,080 --> 00:07:43,980 tuition fees, my living costs to do research on whatever I 129 00:07:43,980 --> 00:07:46,960 wanted to do. There was a second offer from 130 00:07:46,968 --> 00:07:50,392 MIT that was specifically to work in boiling heat transfer, 131 00:07:50,400 --> 00:07:55,001 which was a subject that I I was really interested in. 132 00:07:55,009 --> 00:08:01,960 However, the downsize was that I had to earn my keep as a 133 00:08:01,968 --> 00:08:07,679 teaching assistant, that I'd spend maybe 15, 20 even 20 hours 134 00:08:07,679 --> 00:08:12,000 a week tutoring students probably brighter than me on on 135 00:08:12,000 --> 00:08:14,395 the various courses they'd be taking. 136 00:08:14,403 --> 00:08:19,488 So I knew where my preference was, but I thought I'd go back 137 00:08:19,496 --> 00:08:24,258 and ask her for an appointment with Brian Spalding to confirm 138 00:08:24,266 --> 00:08:28,180 my choice. His secretary graciously gave me 139 00:08:28,180 --> 00:08:33,914 a 5 minute interview slot. I went in and took perhaps a 140 00:08:33,922 --> 00:08:38,898 minute describing my fortunate pair of choices and he said 141 00:08:38,905 --> 00:08:39,909 Princeton. Huh. 142 00:08:39,918 --> 00:08:44,200 Well, Princeton's Princeton. Bob Drake, who's just 143 00:08:44,207 --> 00:08:48,432 co-authored a heat transfer textbook with Professor Eckert. 144 00:08:48,432 --> 00:08:50,640 But you know, from my meetings 145 00:08:50,640 --> 00:08:54,200 with him, Drake hasn't had an original idea in his life. 146 00:08:54,400 --> 00:08:58,040 Go to MIT with Rohsenow and do boiling heat transfer. 147 00:08:58,680 --> 00:09:03,800 This response somewhat deflated me again, but equally I felt I 148 00:09:03,800 --> 00:09:11,440 had to take take his advice and reluctantly declined the 149 00:09:11,440 --> 00:09:15,920 Princeton scholarship. But before I got to write to 150 00:09:16,120 --> 00:09:21,440 MIT, a counter offer came in from their gas turbine 151 00:09:21,440 --> 00:09:25,760 laboratory and the important difference was that they offered 152 00:09:25,760 --> 00:09:31,040 me not a teaching assistantship, but a research assistantship. 153 00:09:31,320 --> 00:09:35,520 This meant that I would do research on one of their 154 00:09:35,520 --> 00:09:38,600 projects, which would also become my thesis. 155 00:09:38,600 --> 00:09:41,560 This was a no brainer for me at any rate. 156 00:09:41,720 --> 00:09:46,880 So I accepted that offer. I declined the heat transfer 157 00:09:46,880 --> 00:09:51,280 offer, and I'm afraid that from there on in my life, boiling 158 00:09:51,280 --> 00:09:55,160 heat transfer had sunk beneath my horizon. 159 00:09:56,920 --> 00:10:02,347 OK, Well, before I talk about my research at MIT and 160 00:10:02,347 --> 00:10:06,346 thereafter, let me just switch back for a minute to Hugh 161 00:10:06,346 --> 00:10:10,320 Kendrick. We were talking one day and I 162 00:10:10,328 --> 00:10:14,864 said to him, hey, I have accepted a postgraduate position 163 00:10:14,864 --> 00:10:19,360 at MIT, where are you going? You'd talked about going to 164 00:10:19,368 --> 00:10:22,400 Princeton or Yale. I suddenly realized I was 165 00:10:22,408 --> 00:10:26,640 talking that someone with the blood draining from his face. 166 00:10:26,648 --> 00:10:31,160 He was literally dumbfounded at what I told him. 167 00:10:31,168 --> 00:10:36,384 Eventually he muttered that he had a he had a graduate 168 00:10:36,392 --> 00:10:40,312 apprenticeship, signed up with Vickers, that he hadn't even 169 00:10:40,320 --> 00:10:42,540 been looking at other opportunities. 170 00:10:42,548 --> 00:10:46,250 He'd go and complete his postgraduate apprenticeship with 171 00:10:46,250 --> 00:10:49,185 them. But then I could almost see the 172 00:10:49,185 --> 00:10:51,275 wheels turning around in his brain. 173 00:10:51,283 --> 00:10:55,855 He said with you could see growing confidence, he said, But 174 00:10:55,855 --> 00:10:59,432 I was under 21 when I signed that contract. 175 00:10:59,440 --> 00:11:04,588 It's not enforceable. And I'm sure MIT will take in a 176 00:11:04,588 --> 00:11:07,648 thicko like you. They'll, they'll find a place 177 00:11:07,648 --> 00:11:09,760 for me. Well, he was right. 178 00:11:09,760 --> 00:11:13,840 They did find a place for him, but by then it was two or three 179 00:11:13,840 --> 00:11:16,640 months after I'd been in touch with them. 180 00:11:16,640 --> 00:11:19,160 They'd allocated all their funding for the year. 181 00:11:19,320 --> 00:11:24,560 However, not to be deterred, he applied to Caltech on the other 182 00:11:24,560 --> 00:11:28,840 side of America and was offered a teaching assistantship. 183 00:11:28,960 --> 00:11:38,560 And that was in August 1961. He and I sailed across in a very 184 00:11:38,560 --> 00:11:41,040 tiny cabin on the Queen Mary together. 185 00:11:41,160 --> 00:11:46,600 Thereafter I caught the train up to MIT, or rather to Boston. 186 00:11:46,600 --> 00:11:51,040 But then Boston is just across the Charles River in Cambridge 187 00:11:51,280 --> 00:11:53,622 and not a just a short distance. 188 00:11:53,630 --> 00:11:58,332 He flew out to Los Angeles and indeed made the rest of his life 189 00:11:58,332 --> 00:12:00,540 in America that. That's that's amazing. 190 00:12:00,548 --> 00:12:03,920 And I I'm more interested and I don't know if other people 191 00:12:03,928 --> 00:12:07,160 hearing this is the comment about sailing across from the 192 00:12:07,168 --> 00:12:10,888 Queen Mary, mainly because now if I go to America and it'll 193 00:12:10,896 --> 00:12:13,464 probably the same for you, you know, it's a flight. 194 00:12:13,472 --> 00:12:15,595 How long did it take on the boat? 195 00:12:15,603 --> 00:12:19,539 It was it was a five-day voyage. It was convenient because we 196 00:12:19,547 --> 00:12:24,285 worked on 25 hour days so that so that we are on the right 197 00:12:24,293 --> 00:12:27,522 time. When we finally arrived in New 198 00:12:27,530 --> 00:12:32,520 York, it was fine, fine sea journeys, a bit boring, you 199 00:12:32,528 --> 00:12:36,545 know, try to amuse oneself. We had a tiny cabin, but we 200 00:12:36,553 --> 00:12:40,114 would sneak through into the, I'm not sure if it was first 201 00:12:40,122 --> 00:12:45,154 class or second class areas and were able to watch films and and 202 00:12:45,154 --> 00:12:49,280 and so on in their luxury, luxury cinemas. 203 00:12:49,288 --> 00:12:54,090 And so that they had on. It was an interesting experience 204 00:12:54,090 --> 00:12:57,630 to do once. The other thing I remember was 205 00:12:57,638 --> 00:13:02,476 that as we got close to New York, you close to? 206 00:13:02,484 --> 00:13:07,572 Yeah, the final destination, the air became more and more humid. 207 00:13:07,572 --> 00:13:09,960 So now I remember a site sailing 208 00:13:09,960 --> 00:13:12,680 past the Statue of Liberty. Of course. 209 00:13:13,160 --> 00:13:19,240 But it was absolutely 9090% humidity by then it it was 210 00:13:19,240 --> 00:13:21,640 difficult. Anyway, that must have been a 211 00:13:21,640 --> 00:13:24,120 moment, though, to see the Statue of Liberty coming to 212 00:13:24,120 --> 00:13:27,920 America with your friend. Yeah, that's a that's a great 213 00:13:27,920 --> 00:13:30,720 point. So what happened then when you 214 00:13:30,720 --> 00:13:34,920 got to to MIT? How did you decide what research 215 00:13:34,920 --> 00:13:39,360 projects to welcome? Well, I was simply checked in 216 00:13:39,360 --> 00:13:47,160 there and they gave me a folder with perhaps 15 research 217 00:13:47,160 --> 00:13:50,000 projects summarized in there, the sufficient amount of 218 00:13:50,000 --> 00:13:53,594 background that one could make a decision, and I knew pretty 219 00:13:53,594 --> 00:13:57,240 much straight away what my choice would be. 220 00:13:57,240 --> 00:14:01,120 At Imperial I'd learned the rudiments of boundary layer 221 00:14:01,120 --> 00:14:04,000 transition. That is to say, a boundary layer 222 00:14:04,000 --> 00:14:08,400 developed initially was laminar, but then when the Reynolds 223 00:14:08,400 --> 00:14:11,840 number got high enough, conditions were right for a 224 00:14:11,840 --> 00:14:16,080 transition to a turbulent boundary layer and there and 225 00:14:16,080 --> 00:14:19,320 turbulent it remained. But this this outline 226 00:14:19,320 --> 00:14:24,040 description that I was reading said that couple of Soviet 227 00:14:24,040 --> 00:14:29,520 scientists had found that in supersonic flow, if a turbulent 228 00:14:29,520 --> 00:14:33,702 boundary layer was passed around what is called a a 229 00:14:33,702 --> 00:14:37,264 Prandtl–Meyer expansion, essentially that's a series of 230 00:14:37,264 --> 00:14:41,816 waves that one might go through in turning a corner and the 231 00:14:41,816 --> 00:14:44,945 flow is through that expansion greatly accelerated. 232 00:14:44,953 --> 00:14:49,864 They found that downstream they had a laminar boundary layer 233 00:14:49,872 --> 00:14:54,560 growing again. So this project outlines said 234 00:14:54,568 --> 00:15:01,020 you will determine whether in subsonic flow a boundary layer 235 00:15:01,028 --> 00:15:05,484 that's strongly accelerated also can revert back to 236 00:15:05,484 --> 00:15:10,862 turbulent—sorry, to laminar and I knew I had to take that it, it 237 00:15:10,862 --> 00:15:13,880 just seemed very a very interesting topic and it 238 00:15:13,880 --> 00:15:18,166 occupied the next three years of my life doing doing research 239 00:15:18,166 --> 00:15:21,680 there. And the outcome was that yes, 240 00:15:21,960 --> 00:15:25,160 you could make a turbulent boundary layer go back to 241 00:15:25,160 --> 00:15:27,560 laminar or at least towards laminar. 242 00:15:27,640 --> 00:15:32,000 I established the acceleration parameter, dimensionless 243 00:15:32,000 --> 00:15:36,200 acceleration parameter that would determine when that would 244 00:15:36,200 --> 00:15:41,760 take place and how roughly how big the parameter was. 245 00:15:41,840 --> 00:15:46,840 I even developed what would be called a simple integral profile 246 00:15:46,840 --> 00:15:50,200 method of predicting the phenomenon based on the idea 247 00:15:50,200 --> 00:15:54,680 that whilst the viscous stresses were greatly increased by the 248 00:15:54,960 --> 00:15:58,760 acceleration, that happened too quickly for the turbulent 249 00:15:58,760 --> 00:16:01,680 stresses to respond. So I just left those as they 250 00:16:01,720 --> 00:16:06,240 were and doing calculations based on that idea pretty much 251 00:16:06,320 --> 00:16:10,360 imitate mimic the experiments that I'd. 252 00:16:10,480 --> 00:16:15,600 OK, so now having finished my work at MIT, what next? 253 00:16:15,600 --> 00:16:20,760 There were there were jobs available for PhD graduates from 254 00:16:20,760 --> 00:16:26,200 MIT offering $14,000 or so a year, but I wasn't eligible to 255 00:16:26,200 --> 00:16:32,280 apply for those. I'd taken a a Fulbright travel 256 00:16:32,280 --> 00:16:36,880 grant to come across to the US, but remember that at the end of 257 00:16:36,880 --> 00:16:40,880 it, I had to go back. So finding a job back in 258 00:16:40,880 --> 00:16:46,560 England, I'd had one or two offers came in for postgraduate 259 00:16:46,720 --> 00:16:50,960 research studentships. Not studentships, postgraduate 260 00:16:51,920 --> 00:16:57,080 employment at government laboratories like the National 261 00:16:57,080 --> 00:17:00,800 Gas Turbine Establishment, central electricity research 262 00:17:00,800 --> 00:17:02,920 labs, and so on. They were. 263 00:17:03,560 --> 00:17:07,079 They weren't uninteresting, but they didn't quite tick all the 264 00:17:07,079 --> 00:17:10,240 boxes. So I decided to write to Brian 265 00:17:10,240 --> 00:17:15,760 Spalding again and wondering whether there was a post and he 266 00:17:15,760 --> 00:17:20,680 said send a CV. So I sent a CV of what I'd 267 00:17:20,920 --> 00:17:25,440 accomplished and I just was working on proofs of the 1st 268 00:17:25,440 --> 00:17:29,800 paper that I'd written and I decided to send those along too 269 00:17:29,800 --> 00:17:32,400 for good measure. It may sound a long time. 270 00:17:32,400 --> 00:17:35,440 Everything was roughly 2 weeks between writing and getting. 271 00:17:35,440 --> 00:17:36,600 I was just. Going to say yeah. 272 00:17:37,240 --> 00:17:38,160 It wasn't. E-mail. 273 00:17:38,720 --> 00:17:41,080 Yeah, yeah, there wasn't the e-mail response. 274 00:17:41,080 --> 00:17:48,720 So anyway, 2 weeks later I I got a note from Spalding saying I 275 00:17:48,720 --> 00:17:53,920 could expect the offer of a lectureship shortly, and sure 276 00:17:53,920 --> 00:17:57,576 enough, two or three days later there was a formal note from 277 00:17:57,576 --> 00:18:02,128 the Registrar's department offering me a position as a 278 00:18:02,128 --> 00:18:06,394 lecturer at the bottom of the lecturer scale, which at the 279 00:18:06,394 --> 00:18:12,520 time was 1400 lbs a year. It didn't compare very well with 280 00:18:12,520 --> 00:18:16,320 what I could have earned in the USA if I'd been eligible, but I 281 00:18:16,320 --> 00:18:20,200 forget this was Imperial College in London. 282 00:18:20,800 --> 00:18:26,480 I was also awarded London Allowance and this was an extra 283 00:18:26,480 --> 00:18:32,520 £60 a year. So I, I accepted like a shot and 284 00:18:32,520 --> 00:18:37,040 my early years at Imperial College were, were were just 285 00:18:37,040 --> 00:18:39,880 delightful. I'd really a very light teaching 286 00:18:39,880 --> 00:18:41,960 load, but an interesting teaching load. 287 00:18:41,960 --> 00:18:43,754 It was a master's level course. 288 00:18:43,762 --> 00:18:46,223 I was asked to take in fluid mechanics. 289 00:18:46,231 --> 00:18:51,860 I had plenty of time free to try and get research students. 290 00:18:51,868 --> 00:18:56,288 Sorry, research funding, but then I mentioned research 291 00:18:56,296 --> 00:18:58,970 students. There was also an important 292 00:18:58,978 --> 00:19:01,023 element there, Brian Spalding said. 293 00:19:01,031 --> 00:19:06,260 Would I take over as my administrative load the process 294 00:19:06,260 --> 00:19:12,255 of research student admissions? It meant that I, I looked at all 295 00:19:12,255 --> 00:19:17,287 the proposals that came in, collected further references if 296 00:19:17,287 --> 00:19:20,488 these weren't provided, distributed the applications to 297 00:19:20,488 --> 00:19:24,836 members of staff that I thought might be interested in taking 298 00:19:24,844 --> 00:19:28,622 them and so on. But naturally I got first sight 299 00:19:28,622 --> 00:19:34,546 of all of the students, and it was from that site that I was 300 00:19:34,554 --> 00:19:40,540 able to recruit both Bill Jones and Kemal Hanjalić the projects 301 00:19:40,540 --> 00:19:45,950 as my first PhD students. Now, viewers may well have heard 302 00:19:45,950 --> 00:19:51,912 of these names because they've they've both gone on to very 303 00:19:51,920 --> 00:19:55,464 successful academic careers and published extensively in 304 00:19:55,472 --> 00:19:58,248 Turbulence Modelling. Can I just ask you a question at 305 00:19:58,248 --> 00:19:59,480 that point? Sure. 306 00:19:59,488 --> 00:20:02,312 How? How did it feel to go back to 307 00:20:02,320 --> 00:20:04,600 London? Because you've gone to Boston, 308 00:20:04,608 --> 00:20:08,460 you've gone to MIT. Was it almost a nice feeling to 309 00:20:08,460 --> 00:20:12,545 come back now as a lecturer, not as a student be teaching? 310 00:20:12,553 --> 00:20:16,130 You know, how did it feel to come back there? 311 00:20:16,138 --> 00:20:19,052 Did it feel like progression with Spalding now as a fellow 312 00:20:19,060 --> 00:20:23,465 member of the academic staff? I don't think I I felt was 313 00:20:23,473 --> 00:20:26,450 terribly conscious of a change of status. 314 00:20:26,458 --> 00:20:30,930 What I was aware of didn't, doesn't relate directly to your 315 00:20:30,930 --> 00:20:36,174 question, but was that some of the research students whom I got 316 00:20:36,174 --> 00:20:41,650 to know a bit while I was doing my final undergraduate 317 00:20:41,650 --> 00:20:45,812 year experimental special tasks, they were still there 318 00:20:45,812 --> 00:20:50,139 working on their pitch. So it very much brought home to 319 00:20:50,139 --> 00:20:54,612 me Spalding his comment that if you want a quick PhD, go 320 00:20:54,620 --> 00:20:56,040 elsewhere. Yeah. 321 00:20:56,048 --> 00:20:57,475 OK. OK. 322 00:20:57,483 --> 00:21:01,280 So you sort of skipped ahead a little bit, almost, yeah. 323 00:21:01,288 --> 00:21:05,909 Yeah. So how so how about, you know, 324 00:21:05,917 --> 00:21:08,920 the research sort of project? How about with with Bill? 325 00:21:08,928 --> 00:21:13,540 Jones, Bill Jones I basically continued PhD research, only we 326 00:21:13,540 --> 00:21:18,612 arranged that he would simply look at one particular 327 00:21:18,620 --> 00:21:22,990 configuration of acceleration and that was flow between 328 00:21:22,998 --> 00:21:26,386 converging planes. What I mean by converging planes 329 00:21:26,386 --> 00:21:29,390 is we measured the boundary layer developing on that surface 330 00:21:29,390 --> 00:21:32,814 and the and the other wall just sloped like this. 331 00:21:32,822 --> 00:21:37,960 It's a very special geometry so far as accelerating boundary 332 00:21:37,968 --> 00:21:42,760 layers is concerned because the thickness of the boundary layer 333 00:21:42,760 --> 00:21:46,384 goes down strictly in proportion to the velocity going 334 00:21:46,384 --> 00:21:48,520 up. So the Reynolds number of the 335 00:21:48,520 --> 00:21:53,120 flow reaches an equilibrium in which it stays constant 336 00:21:53,120 --> 00:21:57,320 thereafter and also the acceleration parameter 337 00:21:57,320 --> 00:22:01,440 determining the reversion to laminar flow that also stays 338 00:22:01,440 --> 00:22:03,760 constant. And what we discovered there 339 00:22:03,760 --> 00:22:08,400 that there wasn't just a single point first of all for various 340 00:22:08,400 --> 00:22:12,440 levels of the acceleration parameter, you actually got a 341 00:22:12,440 --> 00:22:18,400 frozen boundary layer was dissimilar from a turbulent 342 00:22:18,400 --> 00:22:20,960 boundary layer. And this the viscous sub layer 343 00:22:20,960 --> 00:22:24,560 was getting progressively thicker and thicker in what was 344 00:22:24,560 --> 00:22:29,640 still a turbulent flow, a self preserving turbulent flow until 345 00:22:29,640 --> 00:22:32,400 you reached a critical acceleration. 346 00:22:33,440 --> 00:22:37,200 And at that level it you had a complete collapsed to turbulent 347 00:22:37,200 --> 00:22:41,280 flow. So a a lot more came out from 348 00:22:41,480 --> 00:22:45,080 from Bill Jones's study than I had I had achieved. 349 00:22:45,280 --> 00:22:50,400 As for Kemal, we actually looked at a project sponsored by the 350 00:22:50,400 --> 00:22:54,160 Berkeley Nuclear Laboratories. They asked us to determine 351 00:22:54,160 --> 00:22:58,280 whether for flying a duct, whether the position where the 352 00:22:58,280 --> 00:23:03,240 velocity reached its maximum would coincide with where the 353 00:23:03,400 --> 00:23:05,720 turbulent shear stress fell to 0. 354 00:23:06,720 --> 00:23:10,877 They'd been doing tests themselves in flow through an 355 00:23:10,877 --> 00:23:14,360 annulus with a rough inner core tube and a smooth outer 356 00:23:14,360 --> 00:23:19,680 containing tube and had assumed that there was this coincidence 357 00:23:19,680 --> 00:23:24,840 and it gave them very hard to understand values for the 358 00:23:24,840 --> 00:23:29,320 respective shear stresses on the rough and the smooth surface. 359 00:23:29,480 --> 00:23:32,520 Could we sort it out, Judy? We did we he didn't look at an 360 00:23:32,520 --> 00:23:37,800 annulus, we looked at parallel flow between parallel plates and 361 00:23:37,800 --> 00:23:40,000 found there was a huge difference. 362 00:23:40,000 --> 00:23:45,120 That was a thesis very well, very well completed by Kimo. 363 00:23:45,440 --> 00:23:49,920 Now I think the next point I should mention was research that 364 00:23:50,120 --> 00:23:55,484 Brian Spalding and his brilliant student Suhas Patankar had 365 00:23:55,484 --> 00:23:58,970 done. They produced a very efficient, 366 00:23:58,970 --> 00:24:02,445 very easy to use boundary layer code. 367 00:24:02,453 --> 00:24:07,522 The only thing that it was weak on was its model of turbulence. 368 00:24:07,522 --> 00:24:09,480 It it had the mixing length 369 00:24:09,480 --> 00:24:12,388 hypothesis. Prandtl's 1925 paper. 370 00:24:12,388 --> 00:24:19,920 In there, Spalding realised that what he needed was a much more 371 00:24:19,920 --> 00:24:25,480 sophisticated model, one that had individual velocity and 372 00:24:25,480 --> 00:24:29,640 length scales of turbulence available, and not only the 373 00:24:29,640 --> 00:24:35,040 these scales would be affected by diffusion, convection and 374 00:24:35,040 --> 00:24:40,200 various source and sink terms. So he'd obviously chosen the 375 00:24:40,200 --> 00:24:41,920 turbulence energy as one of them. 376 00:24:42,680 --> 00:24:46,440 But then there was the issue of what would you use to get the 377 00:24:46,440 --> 00:24:50,600 turbulent length scale didn't have to be the length scale 378 00:24:50,600 --> 00:24:54,380 itself because you already were solving a transport equation 379 00:24:54,380 --> 00:25:00,000 for the kinetic energy K. So any combination of K with L 380 00:25:00,000 --> 00:25:02,680 would be a possibility. Indeed. 381 00:25:02,680 --> 00:25:07,160 Wolfgang Rodi, Spalding's student adopted the product of 382 00:25:07,160 --> 00:25:12,410 K * L where whereas Brian Spalding himself had an 383 00:25:12,410 --> 00:25:16,728 individual project underway and he chose something he called 384 00:25:16,728 --> 00:25:21,838 W, which basically amounted to K divided by the square of 385 00:25:21,838 --> 00:25:24,852 length scale. OK, well one day Spalding said 386 00:25:24,860 --> 00:25:28,779 to me, hey, would you like to join our group? 387 00:25:28,787 --> 00:25:34,663 I put it to Bill and Kimmo and they thought that sounded a neat 388 00:25:34,663 --> 00:25:37,240 idea. By then we'd started to explore 389 00:25:37,240 --> 00:25:40,360 elaborations of the mixing length hypothesis, but although 390 00:25:40,360 --> 00:25:43,544 we've published several papers on the topic, it clearly wasn't 391 00:25:43,544 --> 00:25:46,652 going to get anywhere. So one thing I should say 392 00:25:46,660 --> 00:25:51,300 though, is that Kemal Hanjalić came from Yugoslavia and 393 00:25:51,308 --> 00:25:54,685 Yugoslavia was a communist country at the time. 394 00:25:54,693 --> 00:25:59,102 Naturally then he had learned Russian in school and that meant 395 00:25:59,102 --> 00:26:03,852 he could, he could have ready access to the Russian literature 396 00:26:03,852 --> 00:26:09,360 and came across a paper by somebody called Davidov, which 397 00:26:09,368 --> 00:26:13,775 advocated in an otherwise totally unusable turbulence 398 00:26:13,783 --> 00:26:18,744 model, but it advocated the use of epsilon, the energy 399 00:26:18,752 --> 00:26:23,750 dissipation rate, OK, of, of turbulence that appealed to us 400 00:26:23,758 --> 00:26:26,880 just physically. It was something that one could 401 00:26:26,880 --> 00:26:31,235 put one's hand on and hopefully one day be able to measure 402 00:26:31,243 --> 00:26:37,281 accurately. And thus that was what we chose. 403 00:26:37,281 --> 00:26:41,040 We produced that, I say we, but 404 00:26:41,280 --> 00:26:45,280 I was sitting back, it was my research students that that were 405 00:26:45,280 --> 00:26:49,400 doing the computation. We produced the version of that 406 00:26:49,800 --> 00:26:55,920 and we're delighted to discover that precisely the same model 407 00:26:56,560 --> 00:27:02,600 enabled us to compute both the free flows, that is to say of of 408 00:27:02,840 --> 00:27:08,003 mixing layer or a plane jet, but with the same model that we 409 00:27:08,003 --> 00:27:11,320 used to compute flow along a flat plate. 410 00:27:11,320 --> 00:27:17,280 Neither of the options that Spalding and Rodi were using 411 00:27:17,920 --> 00:27:22,280 achieved that. So it it was indeed real 412 00:27:22,280 --> 00:27:25,040 progress. And fairly soon thereafter, 413 00:27:25,040 --> 00:27:29,440 Brian Spalding advocated that the group's work should focus on 414 00:27:29,440 --> 00:27:33,640 what he called the k-epsilon model rather than either of the 415 00:27:33,640 --> 00:27:37,760 alternatives. Of course, this model that we 416 00:27:37,800 --> 00:27:42,280 arrived at, the k-epsilon eddy viscosity model, wasn't any good 417 00:27:42,280 --> 00:27:45,600 as it stood for either Bill or Kimo's research. 418 00:27:45,680 --> 00:27:51,080 It needed great extension. For example, in Bill's case, he 419 00:27:51,080 --> 00:27:55,360 had to deal with situations where one started with a 420 00:27:55,360 --> 00:27:59,440 turbulent boundary layer and it went back to or towards lamina. 421 00:27:59,880 --> 00:28:04,480 And so with the inner boundary condition that was being used 422 00:28:04,480 --> 00:28:08,360 next to the wall wasn't strictly a wall boundary condition. 423 00:28:08,520 --> 00:28:11,880 It assumed that across the viscous sub layer there was a 424 00:28:11,880 --> 00:28:16,000 universal velocity profile. And so we didn't need to go 425 00:28:16,000 --> 00:28:19,360 actually to the wall. We'd make our boundary condition 426 00:28:19,640 --> 00:28:23,840 out in the fully turbulent region just a millimetre or so 427 00:28:24,000 --> 00:28:27,440 from the wall. That meant that he had to 428 00:28:27,440 --> 00:28:32,600 develop and extend his model and it it was a major task so that 429 00:28:32,600 --> 00:28:38,480 it included all semi viscous effects that would modify both 430 00:28:38,480 --> 00:28:43,200 kinetic energy that that wasn't that difficult, but also also 431 00:28:43,280 --> 00:28:46,400 energy dissipation rate. And this one was flying blind. 432 00:28:46,400 --> 00:28:48,098 There was no experience in this. 433 00:28:48,106 --> 00:28:53,079 So he, he did a very good job and I'm glad to say that the 434 00:28:53,087 --> 00:28:57,462 paper he and I produced describing it as being cited a 435 00:28:57,470 --> 00:29:03,162 large number of times over 6000. I think likewise for chemo, he 436 00:29:03,170 --> 00:29:08,900 had been asked by sponsors to explain or indeed provide a 437 00:29:08,908 --> 00:29:13,780 model for situations where the position of maximum velocity did 438 00:29:13,780 --> 00:29:17,472 not coincide with the position of 0 shear stress. 439 00:29:17,480 --> 00:29:21,067 But of course with an eddy viscosity model, you can't do 440 00:29:21,075 --> 00:29:23,685 that. They're linked by the definitive 441 00:29:23,685 --> 00:29:27,700 link interlinkage between the mean field velocity gradient and 442 00:29:27,700 --> 00:29:33,376 the turbulent shear stress. So there what we did was provide 443 00:29:33,376 --> 00:29:37,468 an additional independent transport equation for the 444 00:29:37,476 --> 00:29:41,865 turbulent stress. Again, that paper has been 445 00:29:41,873 --> 00:29:47,630 reasonably widely cited. I think both Kimmo and Bill and 446 00:29:47,630 --> 00:29:52,551 I felt at the times, though, that eddy viscosity modelling 447 00:29:52,559 --> 00:29:56,458 wasn't the way to go for most cases. 448 00:29:56,466 --> 00:30:01,440 We felt that you really needed to solve transport equations for 449 00:30:01,440 --> 00:30:05,224 all the Reynolds stresses independently of the velocity 450 00:30:05,232 --> 00:30:08,460 field. And curiously enough, Brian 451 00:30:08,468 --> 00:30:14,640 Spalding was the agent enabling me to pursue that work. 452 00:30:14,648 --> 00:30:19,624 Wolfgang Rodi by then had completed all his computations, 453 00:30:19,624 --> 00:30:26,136 but he was in the throes of trying to write a thesis that 454 00:30:26,144 --> 00:30:30,024 met his supervisors very exacting standards in English. 455 00:30:30,032 --> 00:30:37,098 His scholarship had run out by then and so I was able to hire 456 00:30:37,106 --> 00:30:43,232 him as a a quasi postdoctoral fellow and together we worked on 457 00:30:43,232 --> 00:30:46,320 developing a full stress transport model. 458 00:30:46,328 --> 00:30:52,180 And in the time available that also went very, very well. 459 00:30:52,188 --> 00:30:57,958 We produced a solver that nowadays known as the LRR model: 460 00:30:57,958 --> 00:31:01,383 Launder, Reece and Rodi in alphabetical order. 461 00:31:01,391 --> 00:31:07,069 I'd I didn't so, and that too has been quite widely cited. 462 00:31:07,077 --> 00:31:09,755 I think that's a little bit of an understatement. 463 00:31:09,763 --> 00:31:12,366 I think they're quite widely used. 464 00:31:12,374 --> 00:31:15,880 And maybe you just forward maybe getting on to the next point, 465 00:31:15,888 --> 00:31:19,474 just on this chapter, I guess would it be fair to say for 466 00:31:19,482 --> 00:31:24,085 people to realize that all of the work prior to Spalding's 467 00:31:24,093 --> 00:31:28,807 finite volume code was essentially experimental and you 468 00:31:28,807 --> 00:31:32,578 were mainly deriving relationships from the 469 00:31:32,586 --> 00:31:37,000 experimental data, There wasn't as much coding or numerical 470 00:31:37,008 --> 00:31:39,848 simulation work. Is that correct? 471 00:31:39,856 --> 00:31:45,254 It was I'd say 80 to 90% correct there. 472 00:31:45,262 --> 00:31:50,481 There was work on developing solvers people in the in the the 473 00:31:50,481 --> 00:31:55,668 USA was were on to on to that a group at Los Alamos. 474 00:31:55,676 --> 00:31:59,125 There were others in England. You mustn't forget Peter 475 00:31:59,133 --> 00:32:02,770 Bradshaw also, although he was out and out an experimentalist. 476 00:32:02,770 --> 00:32:04,800 He worked with a group at the 477 00:32:04,800 --> 00:32:08,960 National Physical Laboratory that included some numerical 478 00:32:08,960 --> 00:32:13,680 fluid mechanicians and they developed a scheme usually 479 00:32:13,680 --> 00:32:19,000 called Bradshaw's method that indeed at the 1968 Stanford 480 00:32:19,000 --> 00:32:23,800 conference was rather more successful than Spalding and 481 00:32:23,800 --> 00:32:27,200 Patankar scheme with that had used the mixing length 482 00:32:27,200 --> 00:32:32,160 hypothesis, of course. OK, So the this bringing 483 00:32:32,160 --> 00:32:35,200 together, how did it work in practice at that point? 484 00:32:35,200 --> 00:32:38,800 Was Spalding still essentially running the department? 485 00:32:38,880 --> 00:32:42,480 And and so in some ways was that the slight you still had your 486 00:32:42,480 --> 00:32:46,153 individual research groups, but you were sort of merging in 487 00:32:46,153 --> 00:32:48,640 one, you know, what was that? Because I can imagine he is the 488 00:32:48,640 --> 00:32:52,080 senior professor here. You you're the How did that 489 00:32:52,080 --> 00:32:55,600 relationship at that point? Well, well, Spalding was never a 490 00:32:55,600 --> 00:32:58,640 head of department. He was happy to be head of the 491 00:32:58,840 --> 00:33:04,360 heat transfer section, OK, really the next, and he was 492 00:33:04,360 --> 00:33:09,520 happy doing that. The next major change in my own 493 00:33:09,520 --> 00:33:14,960 career also arose indirectly from Brian Spalding with the 494 00:33:15,040 --> 00:33:18,000 success of this new boundary layer code. 495 00:33:18,200 --> 00:33:21,920 And he had another code that handled recirculating flows too. 496 00:33:23,160 --> 00:33:27,960 There was a lot of interest outside of the university in 497 00:33:28,160 --> 00:33:34,000 getting consultancy help on applying these new schemes to 498 00:33:34,000 --> 00:33:37,840 flows that were of interest to the people in question. 499 00:33:37,960 --> 00:33:41,280 Spalding initially tried handling that internally by 500 00:33:41,280 --> 00:33:46,800 sharing tasks with trusted members of staff and they're 501 00:33:46,840 --> 00:33:50,680 research students. But after a year or so, he he 502 00:33:50,680 --> 00:33:55,080 just felt that wasn't working. He'd find somewhere outside of 503 00:33:55,080 --> 00:33:59,240 the university. So he established an independent 504 00:33:59,240 --> 00:34:04,440 company, Concentration, Heat and Momentum, or CHAM for short. 505 00:34:04,440 --> 00:34:09,320 It established headquarters in Wimbledon and then he had the 506 00:34:09,320 --> 00:34:16,280 enormous task of recruiting staff, completing the various 507 00:34:16,760 --> 00:34:19,280 consultancy requests that were coming in. 508 00:34:19,280 --> 00:34:25,520 And inevitably, even for someone as able as him, the activity 509 00:34:25,520 --> 00:34:30,400 within the university itself, his academic role, they'd they 510 00:34:30,400 --> 00:34:38,400 very much were on hold and this this produced unhappiness with 511 00:34:38,679 --> 00:34:44,560 within the department. Now, as it happened, one of the 512 00:34:45,000 --> 00:34:48,600 Spalding's group had also been made a professor by then. 513 00:34:49,080 --> 00:34:53,400 His name was Jim Whitelaw. He was an experimentalist in 514 00:34:53,400 --> 00:34:57,800 fluid mechanics mainly, but the idea developed that since he 515 00:34:57,800 --> 00:35:02,090 was a professor, he was eligible to basically to lead a group 516 00:35:02,090 --> 00:35:05,200 of staff. Maybe the head of department 517 00:35:05,200 --> 00:35:08,680 would form something called the fluid section, and those who 518 00:35:08,680 --> 00:35:13,680 wanted would transfer their affinity from the heat transfer 519 00:35:13,680 --> 00:35:17,240 section to the fluid section. What was the outcome of that? 520 00:35:17,240 --> 00:35:19,390 You can guess absolute turmoil. 521 00:35:19,398 --> 00:35:24,460 I won't go into the details of the battling, but essentially it 522 00:35:24,460 --> 00:35:30,292 brought research to a halt. Then one day I had a telephone 523 00:35:30,300 --> 00:35:34,070 call from the USA. It was from the head of 524 00:35:34,078 --> 00:35:36,538 department at the University of California, Davis. 525 00:35:36,546 --> 00:35:41,685 He'd been on sabbatical leave within Spalding's Group A couple 526 00:35:41,685 --> 00:35:44,992 of years before. He and I had got to know each 527 00:35:45,000 --> 00:35:48,395 other somewhat. So he called and he said 528 00:35:48,403 --> 00:35:53,580 something like, hey, Brian, I really sorry to hear about the 529 00:35:53,588 --> 00:35:58,182 problems that you guys have got in the department at the moment. 530 00:35:58,182 --> 00:36:00,200 I've been talking with the Dean 531 00:36:00,200 --> 00:36:05,360 and he said to me that if I wanted to offer you a full 532 00:36:05,360 --> 00:36:08,800 professorship to come here to Davis, that was that. 533 00:36:09,480 --> 00:36:13,920 Would you be interested? I heard these words and just 534 00:36:13,920 --> 00:36:17,800 just seemed like someone had I was drowning and that someone 535 00:36:17,800 --> 00:36:21,760 had dropped a lifeline from a helicopter to pull me out of the 536 00:36:23,520 --> 00:36:26,520 So we I said yes without hesitation. 537 00:36:26,520 --> 00:36:32,120 It took maybe well, we didn't get, didn't arrive in Davis 538 00:36:32,120 --> 00:36:39,160 until, well, I think it was 96, the summer of 1976, but we were 539 00:36:39,160 --> 00:36:41,200 very pleased to get there at the time. 540 00:36:41,480 --> 00:36:46,080 What was it, what was it like though with your family, 541 00:36:46,080 --> 00:36:48,120 children? Was that and that was a big 542 00:36:48,120 --> 00:36:52,040 move, you know, big thing to move to the US, to leave London. 543 00:36:52,680 --> 00:36:56,040 Was it an excitement to do it in a way? 544 00:36:56,040 --> 00:37:00,280 Was it a challenge to convince your wife and family to to go 545 00:37:00,280 --> 00:37:04,800 with you? With my, my wife, she's Danish 546 00:37:04,800 --> 00:37:09,194 and moving away from moving away from Europe was a big thing 547 00:37:09,194 --> 00:37:15,000 for her. And I softened that by buying, 548 00:37:15,640 --> 00:37:21,104 buying a property in France. My contract at Davis was for 549 00:37:21,104 --> 00:37:26,142 nine months, that is to say just just just until June and then 550 00:37:26,142 --> 00:37:29,080 you were free completely over the summer. 551 00:37:29,160 --> 00:37:33,800 And I'd have to say also that unlike, unlike Imperial College 552 00:37:33,800 --> 00:37:37,800 that I just come out of staff didn't have administrative work 553 00:37:37,800 --> 00:37:41,440 to do. There was other other workers 554 00:37:41,440 --> 00:37:44,760 that handled all the administration academics were 555 00:37:44,760 --> 00:37:47,320 there just to teach and do research. 556 00:37:47,840 --> 00:37:52,520 So it did make sense having this nine month contract. 557 00:37:52,520 --> 00:38:00,920 So what I agreed we do was in in the summer months, we, we come 558 00:38:01,440 --> 00:38:04,560 to Europe entirely, which would give her more opportunity to 559 00:38:04,560 --> 00:38:08,920 interact with friends and and so on. 560 00:38:09,800 --> 00:38:13,600 Yes, we bought a rundown. We bought a rundown farmhouse 561 00:38:13,600 --> 00:38:21,515 just north of Lyon and in the in the summer I would would drop 562 00:38:21,515 --> 00:38:29,174 in two or three times a week to the École Centrale de Lyon in 563 00:38:29,174 --> 00:38:33,840 enjoying doing research with new colleagues there. 564 00:38:33,840 --> 00:38:38,520 My wife was happy at our our rather dilapidated farmhouse, 565 00:38:38,520 --> 00:38:41,360 but that was that was a different life. 566 00:38:41,360 --> 00:38:49,240 So, so yes, we will make that. But what about UC Davis then? 567 00:38:49,240 --> 00:38:52,200 What was it like there? Well, the department was about 568 00:38:52,200 --> 00:38:54,960 half the size of Imperial College. 569 00:38:55,080 --> 00:38:57,880 Colleges staff did a lot more teaching. 570 00:38:57,880 --> 00:39:00,680 They didn't have administration to do, but they certainly did a 571 00:39:00,680 --> 00:39:02,920 lot more teaching. And some of them treated 572 00:39:02,920 --> 00:39:07,920 research really like a hobby, something that they'd squeeze in 573 00:39:07,920 --> 00:39:11,960 when time allowed. Luckily I struck up a very good 574 00:39:11,960 --> 00:39:16,720 working relationship with two staff there that really they 575 00:39:16,720 --> 00:39:20,560 were experimentalists in various aspects of heat transfer and 576 00:39:20,560 --> 00:39:25,680 they they invited me to join them to add a computational side 577 00:39:25,680 --> 00:39:28,400 to their activity. Probably the main thing though 578 00:39:28,400 --> 00:39:32,680 that I should should mention is that ONERA, which is the French 579 00:39:32,680 --> 00:39:38,640 equivalent of NASA ONERA, sent a post doc who had done his 580 00:39:38,640 --> 00:39:42,760 doctorate in in turbulence modelling to work with me for a 581 00:39:42,760 --> 00:39:44,320 year. I guess they thought that the 582 00:39:44,320 --> 00:39:47,640 interaction of the two of us might be useful. 583 00:39:47,800 --> 00:39:53,280 What he had done was produce an equation, sorry, a turbulence 584 00:39:53,280 --> 00:39:58,262 model not with one length scale, but two effectively length 585 00:39:58,262 --> 00:40:01,960 scale equations that are used in different ways in his model. 586 00:40:02,000 --> 00:40:06,160 I didn't find the model as presented in his thesis exactly 587 00:40:06,160 --> 00:40:09,960 coherent. But we work together and agreed 588 00:40:09,960 --> 00:40:14,360 that what we do would be to develop what's known as a multi 589 00:40:14,360 --> 00:40:18,040 scale model. That is to say, we'd effectively 590 00:40:18,200 --> 00:40:22,920 cut the turbulent spectrum in half with a large scale part in 591 00:40:22,920 --> 00:40:29,680 which eddies would capture mean energy from the from the flow 592 00:40:29,680 --> 00:40:35,880 and a medium and fine scale part that received energy from the 593 00:40:36,320 --> 00:40:39,120 large scale turbulence and finally dissipated it. 594 00:40:39,120 --> 00:40:43,120 That is so we had separate transport equations for the two 595 00:40:43,120 --> 00:40:46,280 parts of turbulence. Davis didn't have a great 596 00:40:46,280 --> 00:40:49,480 computing system. We could only look at simple 597 00:40:49,480 --> 00:40:53,400 flows, which also suited very well the background of the 598 00:40:53,520 --> 00:40:56,840 French visitor. So we looked at sudden 599 00:40:56,880 --> 00:40:58,920 contractions, sudden distortions. 600 00:40:58,920 --> 00:41:02,200 They were just, these were just one-dimensional flows. 601 00:41:02,480 --> 00:41:07,000 You could almost calculate them by hand rather than using the 602 00:41:07,000 --> 00:41:12,080 computer. Well anyway, the outcome of this 603 00:41:12,080 --> 00:41:16,800 was I felt quite spectacular. The simple flows which we did a 604 00:41:16,840 --> 00:41:21,920 so so job on with previous models were now much better 605 00:41:21,920 --> 00:41:25,520 predicted. You got a delayed response when 606 00:41:25,520 --> 00:41:29,000 you put new energy in. You didn't get the dissipation 607 00:41:29,000 --> 00:41:33,160 rate occurring straight away. There was a delayed response and 608 00:41:33,160 --> 00:41:37,120 that mimicked very well the experiments that more or less 609 00:41:37,120 --> 00:41:38,920 brought us to the end of his year. 610 00:41:38,920 --> 00:41:43,640 But I wanted to continue the research to include really the 611 00:41:43,640 --> 00:41:46,760 sorts of flows that in engineering 1 is dealing with 612 00:41:47,080 --> 00:41:51,280 boundary layers, mixing layers, jets, wakes this, that and the 613 00:41:51,320 --> 00:41:53,320 other. But in order to do that, the 614 00:41:53,920 --> 00:41:56,560 Davis system wasn't really suitable. 615 00:41:56,560 --> 00:42:01,000 That is to say, the computer gave too slow turn around for 616 00:42:01,000 --> 00:42:05,120 model development. Also, well, I needed to get 617 00:42:05,640 --> 00:42:08,120 funding if I was to get anyone else. 618 00:42:08,200 --> 00:42:11,800 I couldn't easily recruit research students at Davis. 619 00:42:11,800 --> 00:42:14,640 They didn't seem to want to do research in turbulence 620 00:42:14,640 --> 00:42:19,080 modelling. But Kemal Hanjalić, I knew, was 621 00:42:19,080 --> 00:42:22,280 struggling with the administration that had hit him. 622 00:42:22,280 --> 00:42:27,720 And having got funding from NASA, I invited him to come over 623 00:42:27,720 --> 00:42:32,800 for a year and join me. Now, as I say, the Davis 624 00:42:32,800 --> 00:42:36,440 computing system wasn't up to it, but it wasn't difficult to 625 00:42:36,440 --> 00:42:41,160 obtain computing resources at the the Lawrence Berkeley labs. 626 00:42:41,200 --> 00:42:45,360 Now in those days, we're still talking about the 1970s. 627 00:42:45,360 --> 00:42:47,640 You've you've got computing resources. 628 00:42:48,240 --> 00:42:51,516 Laptops weren't invented and you sure didn't have anything 629 00:42:51,516 --> 00:42:55,600 called remote access. In order to use these, one had 630 00:42:55,600 --> 00:43:00,680 to get in one's car and drive the 60 miles or so down to the 631 00:43:00,680 --> 00:43:05,040 Lawrence Berkeley labs, which had brilliant facilities at the 632 00:43:05,040 --> 00:43:09,320 time. Well, Kimo mainly, but I, I in 633 00:43:09,320 --> 00:43:13,040 joined in as well, worked on this problem for the best part 634 00:43:13,040 --> 00:43:16,720 of a year. And I have to say the outcome 635 00:43:16,720 --> 00:43:22,200 was disappointing. We made one or two improvements 636 00:43:22,200 --> 00:43:25,480 that applied equally to single scale models. 637 00:43:25,480 --> 00:43:29,200 But there was really in you, you're looking at flows like a 638 00:43:29,200 --> 00:43:33,600 mixing layer where so much energy is captured from the mean 639 00:43:33,600 --> 00:43:38,120 flow into the turbulence that the medium and fine scale part 640 00:43:38,120 --> 00:43:40,560 of the spectrum hardly had a look in. 641 00:43:40,560 --> 00:43:46,120 So the results that we got for these flows were only marginally 642 00:43:46,120 --> 00:43:50,600 better than we had obtained with a single scale model. 643 00:43:51,400 --> 00:43:55,520 End of story. I thought it worth mentioning 644 00:43:55,520 --> 00:44:00,840 that because this brings home that what looked like very good 645 00:44:00,840 --> 00:44:04,240 research directions sometimes don't don't work. 646 00:44:05,360 --> 00:44:08,120 Which is, which is a good lesson for people, I guess. 647 00:44:08,120 --> 00:44:11,920 Maybe you're well known for your successes with the k-epsilon 648 00:44:11,920 --> 00:44:14,440 model and the LRR. But I guess if the point is not 649 00:44:14,440 --> 00:44:16,680 all research comes out in success. 650 00:44:16,880 --> 00:44:19,200 But maybe that's the point of research, isn't it? 651 00:44:19,440 --> 00:44:21,280 Some stuff works. Indeed doesn't. 652 00:44:21,280 --> 00:44:23,720 Yes. Well, there was something else 653 00:44:23,720 --> 00:44:28,360 that arose from these trips down to Berkeley, because I got to 654 00:44:28,360 --> 00:44:32,480 know the head of department at the University of California, 655 00:44:32,480 --> 00:44:35,240 Berkeley. And one day he he said to me, 656 00:44:35,400 --> 00:44:39,200 listen, I see you're down here quite regularly. 657 00:44:39,240 --> 00:44:44,240 We need someone to teach turbulence modelling to our our 658 00:44:44,320 --> 00:44:47,880 students here. And there are students asking to 659 00:44:47,880 --> 00:44:52,080 do research in that area, but we really don't have anyone that 660 00:44:52,080 --> 00:44:56,560 could provide that role. What if we, we're in the same 661 00:44:56,560 --> 00:45:02,640 university system as Davis? What if we arrange to hire you 662 00:45:02,640 --> 00:45:04,408 for, let's say, one day a week? 663 00:45:04,416 --> 00:45:06,229 It would be very easy to arrange. 664 00:45:06,237 --> 00:45:11,396 And then when you're down here at the LBL, you could also call 665 00:45:11,396 --> 00:45:15,328 in here at the department, see students, give a lecture or two. 666 00:45:15,328 --> 00:45:17,480 And it seemed to me that that 667 00:45:17,800 --> 00:45:22,080 that would work well. Again, I was delighted at this 668 00:45:22,080 --> 00:45:23,960 suggestion. Couldn't wait to get back to 669 00:45:23,960 --> 00:45:27,880 Davis to break the good news to the head of department so he 670 00:45:27,880 --> 00:45:32,520 could then just sign whatever forms were needed to bring the 671 00:45:32,520 --> 00:45:35,800 change about. I told him and he said 672 00:45:36,280 --> 00:45:39,880 absolutely not. If students know they can work 673 00:45:39,880 --> 00:45:43,040 with you by going to Berkeley, they're not going to come to 674 00:45:43,040 --> 00:45:45,760 Davis. I said to him, Alan, that that 675 00:45:45,760 --> 00:45:48,960 doesn't make sense, it's because they won't come to Davis. 676 00:45:49,200 --> 00:45:52,720 But I can. I can see great opportunities 677 00:45:52,720 --> 00:45:55,520 from recruiting students in in Berkeley. 678 00:45:55,960 --> 00:46:01,080 However, he was implacably opposed and I accepted it. 679 00:46:01,080 --> 00:46:05,080 I liked life in California. It was was a bitter pill, but I 680 00:46:05,080 --> 00:46:09,040 could swallow it. But then not very long later, I 681 00:46:09,040 --> 00:46:14,920 got a phone call from England. It was the principle of UMIST, 682 00:46:14,920 --> 00:46:18,000 that is to say Manchester's Institute of Science and 683 00:46:18,000 --> 00:46:20,800 Technology. He said the head of thermofluids 684 00:46:20,800 --> 00:46:25,200 here has died and we're looking to recruit his replacement. 685 00:46:25,360 --> 00:46:28,000 A few people I've asked have suggested you might be 686 00:46:28,000 --> 00:46:31,480 interested. I disabused him of of that idea. 687 00:46:31,480 --> 00:46:36,560 I said listen, I've only been here for 2 1/2 years, I put in a 688 00:46:36,560 --> 00:46:38,600 lot of effort. I haven't started reaping the 689 00:46:38,600 --> 00:46:43,640 rewards of being here so it's it's too early for me to 690 00:46:43,640 --> 00:46:48,240 consider coming. However, he kept persisting and 691 00:46:48,240 --> 00:46:51,323 in the end I thought I'll take a short back trip back to 692 00:46:51,323 --> 00:46:54,474 England, see my parents, haven't seen them for a while, then 693 00:46:54,474 --> 00:46:58,525 I'll go up to Manchester, have the interview and I'll 694 00:46:58,525 --> 00:47:01,994 politely decline if they offer me a position and life will 695 00:47:01,994 --> 00:47:05,920 return to normal. And that was the plan that I put 696 00:47:05,920 --> 00:47:08,520 into motion. Spent a pleasant weekend with my 697 00:47:08,520 --> 00:47:14,000 parents, travelled to Manchester and had a whole day before the 698 00:47:14,000 --> 00:47:19,360 interview to be exposed to what UMIST could offer. 699 00:47:19,360 --> 00:47:23,280 And I have to say Neil, I was just blown away. 700 00:47:24,200 --> 00:47:28,400 First of all, Thermofluids research had its own building 701 00:47:28,480 --> 00:47:34,960 all through itself. It had 15 academic staff that I 702 00:47:34,960 --> 00:47:40,160 would generally have responsibility for, but not just 703 00:47:40,280 --> 00:47:45,040 academic staff. There must have been at least 30 704 00:47:45,280 --> 00:47:50,240 technician staff and what are called experimental officers to 705 00:47:50,240 --> 00:47:52,680 basically help push research along. 706 00:47:52,920 --> 00:47:57,760 There were around computing staff, of course, at the end. 707 00:47:58,120 --> 00:48:02,240 At this point in time everyone was using card decks and it 708 00:48:02,240 --> 00:48:05,240 needed it, needed quite a lot of support. 709 00:48:05,480 --> 00:48:09,480 But four or five full time positions, That was amazing. 710 00:48:09,480 --> 00:48:13,840 Finally, at the end of the day, I had an interview with the 711 00:48:14,480 --> 00:48:19,080 principal himself. It was a relaxed chat really 712 00:48:19,080 --> 00:48:22,300 over a glass of Sherry I recall. 713 00:48:22,308 --> 00:48:27,032 And he said, of course, if you do come here, there will be 714 00:48:27,040 --> 00:48:30,090 administrative work you won't need to get involved in. 715 00:48:30,098 --> 00:48:36,300 And to make it easier for you, you'll have a dowry appointment. 716 00:48:36,300 --> 00:48:38,760 Meaning that I could choose 717 00:48:38,800 --> 00:48:43,680 anyone that I thought was would going to be helpful to come and 718 00:48:43,680 --> 00:48:46,720 work as a lecturer. No interview, no anything. 719 00:48:46,720 --> 00:48:50,560 It was just me saying I want to have him. 720 00:48:51,240 --> 00:48:56,640 It probably runs counter to to in nowadays, but it didn't at 721 00:48:56,640 --> 00:49:00,960 that time. So at the end of that I had to 722 00:49:00,960 --> 00:49:05,440 say I really wanted the interview the next day to go as 723 00:49:05,440 --> 00:49:07,920 well. And it was a disaster. 724 00:49:08,400 --> 00:49:12,800 I was taken into the room itself where the interview was 725 00:49:12,800 --> 00:49:17,040 conducted. In front of me was a huge Oval 726 00:49:17,040 --> 00:49:20,240 table. I was sat on one side of the 727 00:49:20,240 --> 00:49:27,295 table and ranged all around the other side were the 728 00:49:27,295 --> 00:49:30,496 questioners. There were about 10 or 12 there. 729 00:49:30,496 --> 00:49:31,920 The principal was there. 730 00:49:31,920 --> 00:49:38,280 He introduced me, welcomed me, introduced me to the other. 731 00:49:38,880 --> 00:49:42,760 Other people on the panel think he maybe even asked me a couple 732 00:49:42,760 --> 00:49:44,600 of soft questions which I dealt with. 733 00:49:44,600 --> 00:49:49,400 But then he stepped back and turned me loose to these hyenas 734 00:49:49,600 --> 00:49:54,040 that ravaged me for the next 50 minutes or so. 735 00:49:54,040 --> 00:49:57,320 They were obsessed with something called the Finniston 736 00:49:57,320 --> 00:50:02,320 Report, a report, a government produced report by Lord 737 00:50:02,320 --> 00:50:08,320 Finniston or whoever that looked at potential changes in the 738 00:50:08,480 --> 00:50:10,640 training and development of engineers. 739 00:50:10,640 --> 00:50:14,720 I think by then it had been recognized that Britain didn't 740 00:50:14,720 --> 00:50:18,240 recognize engineers to the extent that someone somewhere 741 00:50:18,240 --> 00:50:22,000 like Germany or Japan did. And of course these these 742 00:50:22,000 --> 00:50:25,960 countries, despite the set back of the Second World War, were 743 00:50:26,040 --> 00:50:30,480 already very advanced and overtaking other European 744 00:50:30,480 --> 00:50:34,120 countries. I honestly, I've been in America 745 00:50:34,120 --> 00:50:38,480 for 2 1/2 years. I didn't have a clue about what 746 00:50:38,480 --> 00:50:43,280 my reactions should be. Likewise, and I could have 747 00:50:43,280 --> 00:50:47,680 anticipated this, there were a lot of questions related to what 748 00:50:47,680 --> 00:50:52,160 management style I would apply if I had the post. 749 00:50:52,800 --> 00:50:54,920 I hadn't been talking about management. 750 00:50:54,920 --> 00:50:59,040 I'd I wanted to do research and do some creative teaching. 751 00:50:59,400 --> 00:51:05,840 So again, my my questions were were very weak. 752 00:51:05,840 --> 00:51:10,840 I felt at the end of an hour they released me to catch the 753 00:51:10,840 --> 00:51:14,680 plane back to California. I remember my wife meeting me 754 00:51:15,240 --> 00:51:19,120 and she was quite excited. She said, well, how was it? 755 00:51:19,120 --> 00:51:23,080 Did it did it meet your hopes? We will we be going there? 756 00:51:24,040 --> 00:51:29,400 And I held my hand up and said the job looked fantastic and I 757 00:51:29,400 --> 00:51:33,240 made it a real terrible job of the interview. 758 00:51:33,720 --> 00:51:37,840 That's the end. So I mentally got back to my 759 00:51:37,840 --> 00:51:41,840 work in Davis. But you know, a week later the 760 00:51:41,840 --> 00:51:44,960 principal phoned again and said the job was mine. 761 00:51:45,520 --> 00:51:50,240 It may have been a a fine decision because usually they 762 00:51:50,240 --> 00:51:53,120 don't take that long in reaching a decision. 763 00:51:53,480 --> 00:52:00,360 But anyway, I did accept the position and well, by the time I 764 00:52:00,720 --> 00:52:05,800 got loose from Davis, there were various entanglements that 765 00:52:05,800 --> 00:52:10,440 slowed my departure. It was spring 1980. 766 00:52:10,640 --> 00:52:13,360 After I'd accepted the appointment, the principal 767 00:52:13,360 --> 00:52:18,080 pointed out that I wouldn't be really responsible for the 768 00:52:18,400 --> 00:52:23,000 engine testing work that my predecessor had specialized in. 769 00:52:23,600 --> 00:52:28,320 He had a right hand man for that work. 770 00:52:28,600 --> 00:52:33,142 And if I was agreeable, he could be promoted to professor to 771 00:52:33,142 --> 00:52:35,960 thus removing me from any responsibility. 772 00:52:36,120 --> 00:52:41,920 And then that's what I very much agreed to and for my dowry 773 00:52:41,920 --> 00:52:45,680 lectureship, I recruited Michael Leschziner. 774 00:52:46,560 --> 00:52:51,000 Michael had been my final PhD student before I left Imperial 775 00:52:51,000 --> 00:52:54,080 College. In interim, he'd gone to work 776 00:52:54,080 --> 00:52:56,120 with Wolfgang Rodi in Karlsruhe. 777 00:52:56,120 --> 00:53:00,760 But when I suggested that he might rejoin me as a lecturer at 778 00:53:00,760 --> 00:53:03,160 Manchester, he was very happy to come. 779 00:53:03,920 --> 00:53:08,480 So I was able to get off to a very good start there because I 780 00:53:08,480 --> 00:53:13,360 could delegate to Mike Leschziner responsibility for 781 00:53:13,800 --> 00:53:18,920 the developing software and generally looking after the 782 00:53:18,920 --> 00:53:23,960 turbulence related work while I got experimental work going. 783 00:53:24,880 --> 00:53:30,120 In fact, I was very lucky in that while I was in Davis I'd 784 00:53:30,120 --> 00:53:35,200 negotiated a contract with the Office of Naval Research, ONR as 785 00:53:35,200 --> 00:53:39,000 it's known. The person behind that was 786 00:53:39,000 --> 00:53:44,760 someone who was very sceptical of CFD, of what CFD codes could 787 00:53:44,760 --> 00:53:48,360 do at that time. In particular, he was interested 788 00:53:48,520 --> 00:53:52,760 in flow through heat exchangers, in particular, flow through a 789 00:53:52,760 --> 00:53:59,160 tube, but in a tube that went in a series of U bends as you get 790 00:53:59,160 --> 00:54:03,080 in a heat exchanger. He said he didn't believe that 791 00:54:03,160 --> 00:54:08,040 current CFD software could accurately predict what heat 792 00:54:08,040 --> 00:54:11,960 transfer coefficients would arise from the complications 793 00:54:11,960 --> 00:54:15,150 that the swirling flow that was generated by these U bends 794 00:54:15,150 --> 00:54:19,880 would create. Fine, I accepted that there was 795 00:54:19,880 --> 00:54:23,480 no problem transferring the contract to Manchester. 796 00:54:24,160 --> 00:54:28,680 So we got under way and after we'd got a very nice set of 797 00:54:29,240 --> 00:54:33,934 results, not just for a circular U bend, circular section U 798 00:54:33,934 --> 00:54:38,840 bend, we actually also produced experimental results for a 799 00:54:38,840 --> 00:54:42,840 square section U bend. Real heat exchangers don't have 800 00:54:42,840 --> 00:54:48,200 a square section pipes, of course, but it was easier to do 801 00:54:48,200 --> 00:54:53,000 the experiments using laser Doppler anemometry going through 802 00:54:53,160 --> 00:54:56,760 a plain surface rather than a circular surface at a time. 803 00:54:56,760 --> 00:54:59,280 It took us a little while to work out how to do that. 804 00:55:00,160 --> 00:55:06,680 Anyway, when we went to apply software to that we've we found 805 00:55:07,200 --> 00:55:11,000 very poor agreement. The contract monitor was 806 00:55:11,000 --> 00:55:13,200 delighted he'd been proved correct. 807 00:55:13,560 --> 00:55:18,120 We carried on as well as we could refining the model. 808 00:55:18,120 --> 00:55:24,880 We found in that case that that taking the calculation all the 809 00:55:24,880 --> 00:55:29,080 way to the wall was highly beneficial. 810 00:55:29,240 --> 00:55:34,814 We at that point just put in the mixing length hypothesis 811 00:55:34,814 --> 00:55:38,120 across the sub layer. Remember this, these were 812 00:55:38,120 --> 00:55:40,720 three-dimensional flow calculations and although 813 00:55:40,720 --> 00:55:43,960 computers were getting more and more capable, the calculation 814 00:55:43,960 --> 00:55:49,040 like flow 3 dimensional flow through a a succession of U 815 00:55:49,040 --> 00:55:51,480 bends stretch the resources available. 816 00:55:51,480 --> 00:55:57,680 OK, well we then went experimentally on to rotating U 817 00:55:57,680 --> 00:56:01,440 bends. Now rotating U bends. 818 00:56:01,960 --> 00:56:07,280 Heat exchangers don't rotate, but this configuration also 819 00:56:07,280 --> 00:56:13,600 arises in tubes 1 millimetre in diameter that are inside gas 820 00:56:13,600 --> 00:56:17,920 turbine blades and because of the pressure there and other 821 00:56:18,160 --> 00:56:21,680 other factors. So for Rolls Royce for several 822 00:56:21,680 --> 00:56:28,880 years we continued looking at the impact that swirl has on in 823 00:56:28,880 --> 00:56:32,800 that particular configuration. I think on the research 824 00:56:33,240 --> 00:56:38,960 computing side rather we'd for my money, I'd more or less run 825 00:56:38,960 --> 00:56:43,320 out of road. The there didn't seem to me much 826 00:56:43,320 --> 00:56:48,160 scope within the stress transport models or algebraic 827 00:56:48,160 --> 00:56:51,240 simplifications thereof that we could exploit. 828 00:56:51,360 --> 00:56:55,560 And then one day in a reflective mode, I recalled a paper that 829 00:56:55,760 --> 00:57:00,120 John Lumley had published in 1978. 830 00:57:01,160 --> 00:57:08,000 He made what I then asserted was a absurd suggestion that 831 00:57:08,400 --> 00:57:13,080 turbulence should comply with what he called the two component 832 00:57:13,080 --> 00:57:16,520 limit. Turbulence, as you know, is 3 833 00:57:16,520 --> 00:57:21,600 dimensional, John said. However, if you have a situation 834 00:57:21,600 --> 00:57:27,065 where the fluctuations just lie in a plane, that is a state 835 00:57:27,065 --> 00:57:31,280 that you should insist on your turbulence model agreeing with. 836 00:57:31,480 --> 00:57:35,160 Indeed, he worked out a parameter that automatically 837 00:57:35,160 --> 00:57:39,760 went to zero whenever you did have that two component state, 838 00:57:39,960 --> 00:57:44,000 and thus it would be a useful parameter to use in one's 839 00:57:44,000 --> 00:57:48,360 turbulence model. Well, in reflecting on this that 840 00:57:48,360 --> 00:57:54,268 day, I suddenly realised that as you went closer and closer to 841 00:57:54,268 --> 00:58:00,160 a wall, the presence of the wall damped out fluctuations normal 842 00:58:00,160 --> 00:58:04,480 to the wall, so that then as you get close enough to the wall, 843 00:58:04,480 --> 00:58:10,680 you indeed did have turbulence that was essentially 2 844 00:58:10,680 --> 00:58:13,600 dimensional. At that instant I went from 845 00:58:13,600 --> 00:58:17,360 being a critic to an enthusiastic supporter. 846 00:58:17,800 --> 00:58:21,120 It was, it really was an overnight night thing. 847 00:58:21,120 --> 00:58:26,720 I had two other students at the time but worked on this. 848 00:58:27,640 --> 00:58:32,520 I'm happy to mention the My name 1 was Dimitri Tselepidakis and 849 00:58:32,520 --> 00:58:36,360 the other was Song Fu. Anyway, they made important 850 00:58:36,360 --> 00:58:40,640 first steps, but then their names aren't linked with the TCL 851 00:58:40,640 --> 00:58:44,640 modelling simply because they were in at the development stage 852 00:58:44,760 --> 00:58:51,240 at the exploiting level. Though Li Xiaoping, another 853 00:58:51,240 --> 00:58:57,240 Chinese student who now works for Fluent in the USA, He 854 00:58:57,240 --> 00:59:02,779 applied the scheme both to flows along a flat plate, but also 855 00:59:02,779 --> 00:59:06,280 to riblet flows in order to reduce the drag. 856 00:59:06,440 --> 00:59:09,680 The results from his computations came out well. 857 00:59:09,680 --> 00:59:15,000 And then especially a little afterwards, Tim Craft looking at 858 00:59:15,000 --> 00:59:18,840 a whole range of turbulent flows found that that this two 859 00:59:18,840 --> 00:59:23,320 component limit, the TCL model as we called it, did a brilliant 860 00:59:23,320 --> 00:59:28,600 job of imitating, mimicking the effects of buoyancy 861 00:59:28,600 --> 00:59:34,000 stratification on the behaviour of the turbulent stresses much 862 00:59:34,000 --> 00:59:38,960 better than the other the earlier LRR model that everyone 863 00:59:38,960 --> 00:59:42,960 still uses. Maybe I can ask a quick 864 00:59:42,960 --> 00:59:45,720 question. Actually on the TCL model, it 865 00:59:45,720 --> 00:59:49,760 strikes me that from my understanding, that has probably 866 00:59:49,760 --> 00:59:53,880 the most, that was the peak of the most sophistication, the 867 00:59:53,880 --> 00:59:57,400 most complexity in a turbulence model in, in terms of the 868 00:59:57,400 --> 00:59:59,148 number of equations, the completeness. 869 00:59:59,156 --> 01:00:03,420 And yet if I look at fluent or, you know, other codes today, 870 01:00:03,428 --> 01:00:06,928 it's probably still the case that people use their LR, you 871 01:00:06,936 --> 01:00:10,499 know, LRR or SSG. Is there a reason you think that 872 01:00:10,499 --> 01:00:12,240 the TCL wasn't adopted more widely? 873 01:00:12,248 --> 01:00:14,704 Was it just too difficult to implement? 874 01:00:14,712 --> 01:00:17,192 Do you have any? Yeah, thoughts. 875 01:00:17,200 --> 01:00:22,823 On that, I think it may well be certainly what you suggest was a 876 01:00:22,823 --> 01:00:25,620 contributor. Maybe maybe I didn't get 877 01:00:25,628 --> 01:00:28,681 involved myself in trying to promote the model. 878 01:00:28,689 --> 01:00:32,781 I don't know. Dave Wilcox I think has has done 879 01:00:32,781 --> 01:00:35,840 a better job in in in advocacy. 880 01:00:35,840 --> 01:00:40,640 Florian Menter's another name that is, is is an advocating 881 01:00:40,640 --> 01:00:42,680 that approach. But well, I don't know. 882 01:00:43,080 --> 01:00:45,342 It's a it's a thing of history. 883 01:00:45,350 --> 01:00:48,680 My philosophy is often, and I just wonder whether that's why 884 01:00:48,688 --> 01:00:53,139 the k-epsilon is still today one of the most widely used is there 885 01:00:53,139 --> 01:00:56,775 is a simplicity is often preferred because it's it's 886 01:00:56,783 --> 01:01:00,349 easier to implement, it's easier to debug, it's easier to get 887 01:01:00,357 --> 01:01:02,824 right. And so sometimes I wonder 888 01:01:02,832 --> 01:01:06,080 whether people are inherently favouring simpler models. 889 01:01:06,088 --> 01:01:09,706 And so the k-epsilon, it's probably an easier model to 890 01:01:09,714 --> 01:01:13,749 implement and test and use than the TCL model that is has more 891 01:01:13,757 --> 01:01:17,226 places to make a mistake. I don't know if that's the 892 01:01:17,234 --> 01:01:21,244 reason, it could be war. I I think as well you, you 893 01:01:21,252 --> 01:01:24,832 touched on it, simplicity is a very key point. 894 01:01:24,840 --> 01:01:31,478 But nowadays 1 is, is looking at a situation where computers have 895 01:01:31,478 --> 01:01:36,757 developed so much that if the k-epsilon model isn't good 896 01:01:36,757 --> 01:01:42,006 enough, perhaps you should use a hybrid k-epsilon-LES approach. 897 01:01:42,006 --> 01:01:46,920 Yeah, and LES is conceptually 898 01:01:46,920 --> 01:01:50,000 quite simple. The k-epsilon model is also 899 01:01:50,000 --> 01:01:53,544 pretty simple. I think people can live with 900 01:01:53,552 --> 01:01:57,760 that package, maybe more easily than solving rather obscure 901 01:01:57,768 --> 01:02:03,160 looking models for the pressure strain hypothesis in the in the 902 01:02:03,160 --> 01:02:06,445 TCL. Yeah, but how about, you know, 903 01:02:06,453 --> 01:02:10,235 Tim Craft, somebody who actually taught me when I was at 904 01:02:10,243 --> 01:02:13,256 university? He went further, right? 905 01:02:13,264 --> 01:02:18,897 You went on to move to look at what I would consider I guess a 906 01:02:18,897 --> 01:02:23,568 blend or a theoretical blend of trying to bring some of the 907 01:02:23,576 --> 01:02:26,743 concepts of anisotropy into simpler models. 908 01:02:26,751 --> 01:02:29,516 So this was the non linear cubic models, right? 909 01:02:29,524 --> 01:02:32,640 That was this the next stage, I guess, of the turbulence 910 01:02:32,640 --> 01:02:35,405 modelling work. Very much so. 911 01:02:35,413 --> 01:02:40,880 Toyota had contacted me wanting to send one of their staff at 912 01:02:40,888 --> 01:02:45,310 the time, Kazuhiko Suga, to work with me, and I'd figured that 913 01:02:45,318 --> 01:02:47,614 they wouldn't be interested in TCL modelling. 914 01:02:47,622 --> 01:02:51,724 It really would be a step too far for this car manufacturer. 915 01:02:51,732 --> 01:02:56,788 But we did agree to look at non linear eddy viscosity models. 916 01:02:56,796 --> 01:03:02,720 There had already been half a dozen schemes that brought in 917 01:03:02,728 --> 01:03:08,375 quadratic terms in addition to the principal linear term of an 918 01:03:08,375 --> 01:03:12,990 eddy viscosity model. We started off looking at those 919 01:03:12,990 --> 01:03:18,480 five schemes that already come forward and what we concluded 920 01:03:18,488 --> 01:03:23,905 was that because these five models were also very different, 921 01:03:23,905 --> 01:03:26,995 different magnitudes of coefficients for different 922 01:03:27,003 --> 01:03:31,855 terms, these models had been designed so that they cope with 923 01:03:31,855 --> 01:03:35,950 one particular class of flow that a linearity viscosity model 924 01:03:35,950 --> 01:03:39,700 didn't get right. We concluded that since the 925 01:03:39,708 --> 01:03:44,660 models weren't anything like the same, we would be wasting our 926 01:03:44,668 --> 01:03:48,823 time to continue research at the quadratic level. 927 01:03:48,831 --> 01:03:54,860 So for the first time then, we we explored modelling at cubic 928 01:03:54,868 --> 01:03:58,492 level. Cubic level of course brought in 929 01:03:58,492 --> 01:04:01,896 many more potential terms, each with empirical coefficients 930 01:04:01,896 --> 01:04:06,040 to tune. But Kazuhiko Suga very patiently 931 01:04:06,040 --> 01:04:11,890 and very thoroughly looked at a whole wide range of flows that 932 01:04:11,890 --> 01:04:16,580 were difficult or couldn't be predicted with a linear eddy 933 01:04:16,588 --> 01:04:22,685 viscosity model, and we ended up with a version that perhaps 934 01:04:22,693 --> 01:04:28,085 wasn't the final word in cubic eddy viscosity modelling at at 935 01:04:28,093 --> 01:04:35,071 least did a hell of a lot better for a large number of flows than 936 01:04:35,071 --> 01:04:40,365 than the quadratic models. So he was content. 937 01:04:40,373 --> 01:04:45,823 I'd say after returning to Japan, he worked for a few years 938 01:04:45,823 --> 01:04:49,306 with Toyota. But then this is interesting. 939 01:04:49,314 --> 01:04:55,086 He got permission from Toyota to move to a university position. 940 01:04:55,094 --> 01:04:59,248 Oh nice. You could you could see that as, 941 01:04:59,248 --> 01:05:04,080 as you've just said Neil, a simplification from from the 942 01:05:04,088 --> 01:05:10,208 pinnacle of TCL modelling. And I guess the next step I took 943 01:05:10,208 --> 01:05:12,400 in research went a step further. 944 01:05:12,760 --> 01:05:19,400 What really bugged me was that still in commercial CFD, people 945 01:05:20,000 --> 01:05:26,454 so frequently use wall functions based on this old idea and 946 01:05:26,454 --> 01:05:32,116 very limited, a very limited idea that the near wall 947 01:05:32,116 --> 01:05:33,655 velocity profile was universal. 948 01:05:33,663 --> 01:05:38,175 What could we do with that? Well, I worked on this with Tim 949 01:05:38,175 --> 01:05:42,620 Craft, who had become a lecturer at the time, and Hector 950 01:05:42,620 --> 01:05:46,459 Iacovides, an earlier student that had worked on flow around 951 01:05:46,467 --> 01:05:49,309 bends with me, who was now a professor. 952 01:05:49,317 --> 01:05:55,952 In fact, we decided we'd figure out a better way of building a 953 01:05:55,960 --> 01:06:00,135 wall function. Indeed, we worked on 2 schemes. 954 01:06:00,135 --> 01:06:02,480 There was a analytical approach. 955 01:06:02,480 --> 01:06:08,280 We took the view that although the near wall velocity profile 956 01:06:08,560 --> 01:06:14,040 wasn't universal in most of the flows that one would want to 957 01:06:14,040 --> 01:06:19,880 look at, maybe the turbulent Eddy viscosity would be much 958 01:06:19,880 --> 01:06:23,760 more nearly universal, Particularly as we agreed to 959 01:06:23,760 --> 01:06:28,960 allow the viscous sub layer where there was 0 turbulent 960 01:06:28,960 --> 01:06:31,400 mixing. According to our model, we could 961 01:06:31,400 --> 01:06:36,520 vary that in thickness depending upon the gradient of shear 962 01:06:36,520 --> 01:06:42,640 stress across the layer. Well, this we worked on with a 963 01:06:42,640 --> 01:06:46,160 Russian student. My only Russian student, Aleksey 964 01:06:46,160 --> 01:06:51,120 Gerasimov, was just fun to work with and I'm glad to say that 965 01:06:51,160 --> 01:06:56,520 the model he came up with, what was called the analytical wall 966 01:06:56,520 --> 01:07:01,560 function, did pretty well. It's widely used at Manchester 967 01:07:01,560 --> 01:07:06,120 now and and is at least incorporated in some of the 968 01:07:06,120 --> 01:07:10,960 commercial software. It's a lot, lot better than log 969 01:07:10,960 --> 01:07:16,760 law wall functions. They should be made illegal, but 970 01:07:16,760 --> 01:07:21,560 we also developed a second scheme in situations where you 971 01:07:21,560 --> 01:07:25,640 have the velocity vector changing direction across the 972 01:07:25,640 --> 01:07:30,160 viscous layer, such as a rises in those flow around U bends, 973 01:07:30,160 --> 01:07:33,560 very tight U bends. We developed a numerical scheme. 974 01:07:33,560 --> 01:07:38,680 I I won't attempt to go into any detail on that, but the key 975 01:07:38,680 --> 01:07:43,400 thing was that just as in a boundary layer solver, if you 976 01:07:43,400 --> 01:07:48,920 think of a but 2D boundary layer solver, you treat the static 977 01:07:48,920 --> 01:07:52,880 pressure as though it is uniform across the layer. 978 01:07:53,000 --> 01:07:58,240 We applied the same slight approximation, but only to the 979 01:07:58,840 --> 01:08:02,040 very near wall region covered by the wall function. 980 01:08:02,040 --> 01:08:05,400 And I won't go into the details. 981 01:08:05,408 --> 01:08:09,060 Indeed I've forgotten some of the details by now. 982 01:08:09,068 --> 01:08:13,260 It's a while ago. But that permitted us to very 983 01:08:13,268 --> 01:08:17,475 much reduce the cost of using quite advanced models. 984 01:08:17,483 --> 01:08:21,240 I mean, Bill Jones's low- Reynolds-number model could be 985 01:08:21,247 --> 01:08:26,832 put in or or any any other other model and it reduced the 986 01:08:26,840 --> 01:08:30,840 computing time by something between 80% and 90%. 987 01:08:30,848 --> 01:08:36,386 So it represents a huge saving. I should just maybe add a point 988 01:08:36,386 --> 01:08:40,759 and maybe you're not even aware of this, that I have seen a 989 01:08:40,767 --> 01:08:45,439 resurgence of the interest in the work that you did with Tim 990 01:08:45,448 --> 01:08:49,291 and Hector for advanced wall functions in the context 991 01:08:49,299 --> 01:08:53,319 actually of LES. That now because there is a 992 01:08:53,328 --> 01:08:58,180 resurgence of interest in the wall-modelled LES, there is a 993 01:08:58,188 --> 01:09:03,660 realization that the model used to approximate the near wall 994 01:09:03,667 --> 01:09:07,000 behaviour. Can we look at more advanced 995 01:09:07,008 --> 01:09:10,301 ways of calculating that? And I I've seen now people 996 01:09:10,310 --> 01:09:14,323 reference and look at that work not in the context of RANS, but 997 01:09:14,323 --> 01:09:16,700 actually in the context of LES, which I thought is quite 998 01:09:16,707 --> 01:09:19,975 interesting how work that was done, you know, 30 years ago is 999 01:09:19,975 --> 01:09:24,479 now being re looked at in a in a different context, but still. 1000 01:09:24,479 --> 01:09:24,920 Yeah. 1001 01:09:25,279 --> 01:09:27,960 So anyway, just a. Thank you Neil. 1002 01:09:27,960 --> 01:09:31,200 I, I was not aware of this resurgence of interest. 1003 01:09:31,200 --> 01:09:37,640 Long may it continue, but I, I have to say that that having got 1004 01:09:37,640 --> 01:09:44,520 that far, what I found was that turbulence models was looked on 1005 01:09:44,560 --> 01:09:47,319 as what might be called a mature subject. 1006 01:09:48,000 --> 01:09:52,080 People weren't interested in looking for radical changes in 1007 01:09:52,080 --> 01:09:55,400 in modelling. It was applications that were 1008 01:09:55,400 --> 01:09:59,240 very much to the fore for I didn't feel that that was what I 1009 01:09:59,240 --> 01:10:06,320 was perhaps best at at doing, though I, I did, I did supervise 1010 01:10:06,320 --> 01:10:11,800 one PhD student by a friend of yours, Alastair, and that was 1011 01:10:11,800 --> 01:10:14,280 very successful. He was looking at in-line tube 1012 01:10:14,280 --> 01:10:16,760 banks and it was quite extraordinary. 1013 01:10:16,760 --> 01:10:22,440 We found the the sorts of deviations from from going 1014 01:10:22,440 --> 01:10:26,584 straight through an in-line tube-bank heat exchanger 1015 01:10:26,584 --> 01:10:31,100 tended to develop a diagonal path through that. 1016 01:10:31,108 --> 01:10:33,744 I should maybe just interrupt you slightly. 1017 01:10:33,752 --> 01:10:37,320 I, I was mean to say it before, but maybe this is a good time to 1018 01:10:37,320 --> 01:10:40,525 say it that almost as a, an observation or to people 1019 01:10:40,533 --> 01:10:43,761 listening that I studied at Manchester University with 1020 01:10:43,769 --> 01:10:46,739 Alastair as well. And we did our undergraduate 1021 01:10:46,739 --> 01:10:50,775 degrees and, and you taught us and so did Tim Craft and so did 1022 01:10:50,775 --> 01:10:52,040 Hector Iacovides and these folk. 1023 01:10:52,040 --> 01:10:56,697 Not Michael, I guess Michael, I think moved to Imperial then 1024 01:10:56,697 --> 01:10:59,844 and but we didn't know who you were. 1025 01:10:59,852 --> 01:11:02,376 We didn't know who Tim was. We didn't know. 1026 01:11:02,384 --> 01:11:05,757 And I, I know Alastair and I talk about this now that we feel 1027 01:11:05,757 --> 01:11:07,698 bad in a way. And I'm sure others are like 1028 01:11:07,706 --> 01:11:09,320 that. They see some professor come in, 1029 01:11:09,320 --> 01:11:12,050 teach a subject, second year something and you all, you don't 1030 01:11:12,050 --> 01:11:14,670 mean to be disrespectful, but you know, you don't know who 1031 01:11:14,678 --> 01:11:16,478 they are. And it's only later you look 1032 01:11:16,486 --> 01:11:20,490 back and think, oh wow, I was, you know, taught by this person 1033 01:11:20,490 --> 01:11:23,879 or supervised. And I, I wonder, I think you 1034 01:11:23,887 --> 01:11:27,029 mentioned to me this happens sometimes that people come or 1035 01:11:27,037 --> 01:11:29,570 want to take a selfie with you or something. 1036 01:11:29,578 --> 01:11:33,663 Because now with social media, I guess people know who people are 1037 01:11:33,663 --> 01:11:35,413 more. But at the time we had no clue. 1038 01:11:35,413 --> 01:11:36,680 And it was an honour to be 1039 01:11:36,680 --> 01:11:38,720 taught by you. But I feel guilty that we didn't 1040 01:11:38,720 --> 01:11:42,480 say it at the time. Well, I have no thoughts on 1041 01:11:42,480 --> 01:11:44,360 that. Indeed you mentioned people 1042 01:11:44,360 --> 01:11:46,240 coming in to have selfies with me. 1043 01:11:46,480 --> 01:11:50,200 I I was just amazed and thought they were slightly crazy but 1044 01:11:50,920 --> 01:11:55,560 complied with their request to to be photographed. 1045 01:11:56,600 --> 01:12:01,880 Well, I suppose the one of the reasons that slightly steered me 1046 01:12:01,880 --> 01:12:08,440 away from continuing too deeply in basically CFD applications. 1047 01:12:08,440 --> 01:12:12,440 Much as there were very interesting things to explore, I 1048 01:12:12,840 --> 01:12:16,680 would in fact mention another application case. 1049 01:12:16,680 --> 01:12:21,120 We were looking at trailing vortices behind a wing. 1050 01:12:21,680 --> 01:12:24,880 There was a good set of experimental data, and we 1051 01:12:24,880 --> 01:12:30,000 computed that with a Eddy viscosity model, and as we knew 1052 01:12:30,000 --> 01:12:33,880 it would, the swirl died out far too quickly. 1053 01:12:34,320 --> 01:12:38,097 Now the trailing vortices behind a wing in practice are known 1054 01:12:38,097 --> 01:12:42,760 to persist a long time, so that a big aircraft landing at an 1055 01:12:42,760 --> 01:12:48,480 airport would make it really dangerous for a plane coming in 1056 01:12:48,640 --> 01:12:53,080 a minute or so after them on the same track because the swirl 1057 01:12:53,080 --> 01:12:59,000 persisted and persisted. The Eddie Viscosity model said 1058 01:12:59,000 --> 01:13:02,480 don't worry folks, the swirl dies out very quickly. 1059 01:13:02,480 --> 01:13:06,560 But experiments that Peter Bradshaw had done said, no, it 1060 01:13:06,560 --> 01:13:09,280 doesn't die out. It hangs around an awful long 1061 01:13:09,280 --> 01:13:12,560 time. And I'm happy to say that one of 1062 01:13:12,560 --> 01:13:19,080 the applications that I did with Tim Craft at Manchester was to 1063 01:13:19,080 --> 01:13:21,383 look at this with the TCL model. 1064 01:13:21,391 --> 01:13:25,742 And the TCL model was the only one of four schemes that we 1065 01:13:25,750 --> 01:13:29,912 looked at that in any way mimicked the observed 1066 01:13:29,920 --> 01:13:33,705 experimental behavior. I will just add if you don't 1067 01:13:33,713 --> 01:13:36,600 mind a quick again interesting thing that you may or may not be 1068 01:13:36,600 --> 01:13:38,729 aware of and it's probably one of the reasons that Alastair 1069 01:13:38,737 --> 01:13:42,565 works where he does now at the Formula One team, is that use 1070 01:13:42,573 --> 01:13:45,370 case of RANS turbulence modelling for highly vortex 1071 01:13:45,370 --> 01:13:48,612 driven flows is a key use case in Formula One. 1072 01:13:48,620 --> 01:13:52,730 And actually from when I was working and even recently, I 1073 01:13:52,738 --> 01:13:55,884 still see people looking at nonlinear eddy- viscosity models 1074 01:13:55,884 --> 01:13:59,830 and Reynolds-stress models because the flow around a 1075 01:13:59,830 --> 01:14:03,502 Formula One car is driven by 10s or more of individual 1076 01:14:03,502 --> 01:14:07,884 vortices and they find that they need these non linear terms 1077 01:14:07,884 --> 01:14:12,302 to capture it more nicely. So even today, I think that's 1078 01:14:12,310 --> 01:14:16,260 one of the industries that still is very much into turbulence 1079 01:14:16,260 --> 01:14:18,822 modelling. So yeah, yeah, that's that, that 1080 01:14:18,822 --> 01:14:23,208 wing tip exact test case I've seen used internally at in F1 1081 01:14:23,216 --> 01:14:25,400 teams to investigate different turbulence models. 1082 01:14:25,408 --> 01:14:27,200 So. Gosh. 1083 01:14:27,208 --> 01:14:31,780 Well, thank you. I hadn't supposed that that 1084 01:14:31,788 --> 01:14:36,664 ground based object like a racing car would encounter the 1085 01:14:36,672 --> 01:14:40,460 same problems. Yeah, that's because all the 1086 01:14:40,468 --> 01:14:43,814 wings, they generate the tips and there's so many of those 1087 01:14:43,822 --> 01:14:46,468 individual wings, there's, there's probably 40 different 1088 01:14:46,476 --> 01:14:49,990 vortices I guess on the car all touching each other. 1089 01:14:49,998 --> 01:14:52,405 You need a pretty fine. It's a result, yes. 1090 01:14:52,413 --> 01:14:55,116 Well, yeah, there's like a billion grid points nowadays, so 1091 01:14:55,116 --> 01:14:58,260 things have moved on. But yes, sorry I interrupted 1092 01:14:58,268 --> 01:15:00,910 you. I think you were talking more 1093 01:15:00,918 --> 01:15:04,200 about the Osborne Reynolds and and some of that work. 1094 01:15:04,208 --> 01:15:07,565 Yeah, well, I've really reached the end of the end of the line. 1095 01:15:07,565 --> 01:15:09,280 I thought I was also getting a 1096 01:15:09,280 --> 01:15:14,157 bit old and so I was interested that the Royal Society 1097 01:15:14,157 --> 01:15:19,728 announced that they had the original copies of referees 1098 01:15:19,728 --> 01:15:22,520 reports from some of the early papers. 1099 01:15:22,528 --> 01:15:25,880 Now I was quite interested in Osborne Reynolds, of course. 1100 01:15:25,888 --> 01:15:29,480 He'd been a professor at Manchester, forerunner of 1101 01:15:29,488 --> 01:15:34,380 Manchester University, but also he his paper on Reynolds 1102 01:15:34,388 --> 01:15:37,200 averaging. Whilst it didn't actually get 1103 01:15:37,208 --> 01:15:41,358 into the problems of modelling the Reynolds stresses, at least 1104 01:15:41,358 --> 01:15:45,052 it marked a starting point for the subject that I'd been 1105 01:15:45,060 --> 01:15:47,610 involved in for most of my professional career. 1106 01:15:47,618 --> 01:15:52,352 So he asked whether they had copies of Reynolds papers. 1107 01:15:52,360 --> 01:15:57,658 Judy travelled down to London to see them. 1108 01:15:57,666 --> 01:16:02,275 And again was I was just blown away. 1109 01:16:02,283 --> 01:16:06,980 I was blown away in much the same way that first coming to 1110 01:16:06,988 --> 01:16:12,992 UMIST had had done for me. You got communications from 1111 01:16:13,000 --> 01:16:19,245 Horace Lamb, George Stokes, from Reynolds of course, really from 1112 01:16:19,245 --> 01:16:23,820 from all of these people that were involved in refereeing or 1113 01:16:23,828 --> 01:16:28,928 acting as editor of of the manuscripts that Reynolds had 1114 01:16:28,936 --> 01:16:33,478 submitted. So I said, well, can I have a 1115 01:16:33,486 --> 01:16:35,686 copy of these? Yes. 1116 01:16:35,694 --> 01:16:39,624 The Royal Society said no, I could not Xerox them. 1117 01:16:39,632 --> 01:16:44,250 They wouldn't allow me to Xerox them, but they could arrange for 1118 01:16:44,250 --> 01:16:49,228 their photographer to come in and copy them for me, which they 1119 01:16:49,228 --> 01:16:53,417 duly did. I was charged 75 lbs for the 1120 01:16:53,425 --> 01:16:57,752 privilege which I the other day I looked up as to what that 1121 01:16:57,760 --> 01:17:01,374 would be today. It was a little over £200 but I 1122 01:17:01,374 --> 01:17:06,040 happily paid that in order to get get the photographs and 1123 01:17:06,048 --> 01:17:13,320 subsequently then I wrote a paper on how despite crushing 1124 01:17:13,328 --> 01:17:18,860 referees reports, nevertheless Osborne Reynolds did publish his 1125 01:17:18,860 --> 01:17:24,599 paper on Reynolds averaging. Well that's a lesson on even the 1126 01:17:24,599 --> 01:17:28,440 famous and well known people like Osborne Reynolds can get 1127 01:17:28,448 --> 01:17:31,896 very harsh reviews, so maybe that's reassuring for everybody 1128 01:17:31,896 --> 01:17:34,806 else who probably also has had those. 1129 01:17:34,814 --> 01:17:39,460 I'm sure it wouldn't have been published except that the 1130 01:17:39,468 --> 01:17:44,500 earlier paper that he'd published in which he discovered 1131 01:17:44,500 --> 01:17:48,928 that transition occurred at a particular of what we call today 1132 01:17:48,928 --> 01:17:50,587 a particular critical Reynolds number. 1133 01:17:50,595 --> 01:17:53,984 If he hadn't published that experimental paper first because 1134 01:17:53,984 --> 01:17:58,272 the referees were saying things like the previous work he 1135 01:17:58,272 --> 01:18:01,304 did was very good. Maybe there's something in this, 1136 01:18:01,304 --> 01:18:05,794 but I can't see it that that was probably the reason it got 1137 01:18:05,802 --> 01:18:07,920 published. Of course, having having 1138 01:18:07,928 --> 01:18:11,988 published one paper, you get a request from conference 1139 01:18:11,996 --> 01:18:17,656 organiser, could I give a paper just like that one? 1140 01:18:17,664 --> 01:18:23,720 And of course my response is I couldn't bear to give the same 1141 01:18:23,728 --> 01:18:27,082 paper twice. But what I discovered was that 1142 01:18:27,090 --> 01:18:31,615 at Manchester they had all of the archive material from 1143 01:18:31,623 --> 01:18:34,480 Reynolds applying for a his chair. 1144 01:18:34,488 --> 01:18:38,400 Retired colleague of mine had also been following up on 1145 01:18:38,408 --> 01:18:41,646 Reynolds's life, a person called Derek Jackson. 1146 01:18:41,654 --> 01:18:48,342 And so we collaborated in producing a much fuller paper on 1147 01:18:48,342 --> 01:18:54,000 Reynolds's life as a whole that was was published in a book 1148 01:18:54,000 --> 01:18:56,596 that I forget. Let me find the name of the book 1149 01:18:56,596 --> 01:18:58,620 here. It is A Voyage Through 1150 01:18:58,628 --> 01:19:02,574 Turbulence that was edited by Keith Moffatt and others from 1151 01:19:02,582 --> 01:19:05,520 Cambridge. Anyway, I don't know whether it 1152 01:19:05,520 --> 01:19:09,978 was connected with these papers when colleagues in the Royal 1153 01:19:09,986 --> 01:19:13,958 Society died. It's the practice of the Royal 1154 01:19:13,966 --> 01:19:19,990 Society to produce a 25 or 30 page memoir on their lives, 1155 01:19:19,998 --> 01:19:25,349 their achievements. So I was when my friend Jim 1156 01:19:25,357 --> 01:19:29,508 Whitelaw passed away, I was asked to write his. 1157 01:19:29,516 --> 01:19:34,060 And you know, that is, that is really an interesting branch of 1158 01:19:34,060 --> 01:19:39,080 research that was quite like looking at looking at Osborne 1159 01:19:39,088 --> 01:19:44,521 Reynolds's past life because Jim had produced a huge catalog of 1160 01:19:44,529 --> 01:19:48,350 what he'd done. And so as a matter of sifting 1161 01:19:48,358 --> 01:19:51,892 through that, pulling out the aspects that would be 1162 01:19:51,900 --> 01:19:55,335 particularly important to stress, plus any personal views 1163 01:19:55,335 --> 01:19:59,452 I might have. So that started following Brian 1164 01:19:59,452 --> 01:20:02,150 Spalding's death. I I also. 1165 01:20:02,158 --> 01:20:06,780 Contributed to his memoir and now most recently, Peter 1166 01:20:06,788 --> 01:20:11,760 Bradshaw, who died last year. I've written his article. 1167 01:20:11,768 --> 01:20:18,455 So life in my retirement hasn't, hasn't by any means been without 1168 01:20:18,455 --> 01:20:23,160 involvement with turbulence. Brian Spalding, When we started 1169 01:20:23,160 --> 01:20:27,210 off in the earlier discussions about the beginning of your 1170 01:20:27,218 --> 01:20:31,762 career, you had the initial chat about where to study, then the 1171 01:20:31,770 --> 01:20:36,252 offer of coming back to Imperial and obviously then there was a a 1172 01:20:36,252 --> 01:20:39,908 sort of falling out and then leaving to UC Davis. 1173 01:20:39,916 --> 01:20:44,118 What was your relationship like later on in in your career? 1174 01:20:44,126 --> 01:20:47,480 Did you stay in touch? Did you? 1175 01:20:47,488 --> 01:20:50,260 Yeah. How was your relationship when 1176 01:20:50,268 --> 01:20:54,920 you came back to UMIST? Well, when I came back to UMIST 1177 01:20:54,920 --> 01:20:57,215 it was. It was static. 1178 01:20:57,223 --> 01:21:03,259 Nothing had moved by then, but then when in 1994 received a 1179 01:21:03,267 --> 01:21:09,822 note notice that I was to be admitted to the Royal Society, I 1180 01:21:09,822 --> 01:21:14,320 got a warm, very brief but brief brevity is certainly 1181 01:21:14,320 --> 01:21:18,140 Spalding style. A brief note congratulating me. 1182 01:21:18,140 --> 01:21:20,640 I didn't respond to that. 1183 01:21:20,640 --> 01:21:23,720 I think in as generous a way as I could. 1184 01:21:23,720 --> 01:21:29,920 We, we might have developed much closer linkages again when he 1185 01:21:29,920 --> 01:21:34,640 was when he was reaching his 90th birthday, there was a, a 1186 01:21:34,640 --> 01:21:38,820 celebration for him and I, I was distantly involved in helping 1187 01:21:38,820 --> 01:21:43,320 to organize that. And of course we met there and 1188 01:21:43,600 --> 01:21:46,560 exchanged pleasantries between us. 1189 01:21:46,560 --> 01:21:50,120 I I guess there was still slightly a feeling of the 1190 01:21:50,120 --> 01:21:52,600 tension. I think the last time I saw him 1191 01:21:52,960 --> 01:21:57,520 was when I was giving, We were both giving, in fact invited 1192 01:21:57,520 --> 01:22:00,200 lectures at a meeting in Sarajevo. 1193 01:22:00,200 --> 01:22:05,040 I would by then had worked into the area of climate change and 1194 01:22:05,360 --> 01:22:09,560 that was the topic of my lecture, really suggesting how 1195 01:22:09,560 --> 01:22:14,397 engineers with their experience and now of being able to 1196 01:22:14,397 --> 01:22:19,490 predict turbulent swirling flows could actually contribute to 1197 01:22:19,490 --> 01:22:24,614 the modelling of hurricanes, perhaps devising methods to 1198 01:22:24,614 --> 01:22:27,914 diminish the intensity of a hurricane. 1199 01:22:27,922 --> 01:22:33,464 I think the paper went over the heads of those there. 1200 01:22:33,472 --> 01:22:39,602 There was, I think, no question at all from the floor except 1201 01:22:39,610 --> 01:22:43,552 from Brian Spalding. He was sat in the front row of 1202 01:22:43,560 --> 01:22:46,903 the audience. We did mend our fences between 1203 01:22:46,911 --> 01:22:50,072 us. I didn't do as much as I should 1204 01:22:50,072 --> 01:22:55,267 have done, and I'm sorry, but I hope in contributing to his bio 1205 01:22:55,267 --> 01:23:01,582 memoir, I'm forgiven. My other question, I'm kind of 1206 01:23:01,590 --> 01:23:05,400 see how you saw it from the outside. 1207 01:23:05,408 --> 01:23:10,659 Often the turbulent modelling, particularly in a RANS context, 1208 01:23:10,659 --> 01:23:15,070 it's often seen as a almost a competition between Philippe 1209 01:23:15,070 --> 01:23:20,350 Spalart, you know, Florian Menter yourself, Wilcox, how 1210 01:23:20,350 --> 01:23:23,507 much did you see it that way in the night? 1211 01:23:23,515 --> 01:23:26,962 And I guess you had done it a little bit before, but in the 1212 01:23:26,970 --> 01:23:31,300 Seventies, 80s and 90s, were you aware of this almost papers 1213 01:23:31,308 --> 01:23:34,220 always comparing the different turbulence modelling approaches 1214 01:23:34,220 --> 01:23:38,344 and which one was better? Was that something you aware of 1215 01:23:38,344 --> 01:23:42,255 or was it not? Yeah, part of your thinking. 1216 01:23:42,263 --> 01:23:48,380 I guess one could say I was dimly aware of, but I was so 1217 01:23:48,388 --> 01:23:52,832 interested, you might say obsessed with tackling the 1218 01:23:52,840 --> 01:23:58,755 problems in research that I felt I needed to deal with. 1219 01:23:58,763 --> 01:24:03,330 I I wasn't looking over my shoulder all the time. 1220 01:24:03,338 --> 01:24:07,575 Looking over one shoulder in implies people are behind me. 1221 01:24:07,583 --> 01:24:11,190 Maybe some some of. Do you think it's ultimately 1222 01:24:11,198 --> 01:24:15,440 though, just a matter of where you studied and what your 1223 01:24:15,448 --> 01:24:20,335 funding was, new use cases, If you're in at Stanford and you're 1224 01:24:20,335 --> 01:24:24,176 in the US, you have close links with NASA and the 1225 01:24:24,176 --> 01:24:26,783 aerospace industry, you're going to be solving airplane 1226 01:24:26,783 --> 01:24:29,520 cases. And so I wonder whether that 1227 01:24:29,520 --> 01:24:31,640 motivated the Spalart–Allmaras and those. 1228 01:24:31,640 --> 01:24:37,600 And if you had, instead of being funded by, you know, Boeing or 1229 01:24:37,600 --> 01:24:41,720 or Rolls Royce, sorry, or Airbus, or would you have 1230 01:24:41,720 --> 01:24:45,360 naturally pivoted and maybe come up with a slightly different 1231 01:24:45,360 --> 01:24:49,120 terms model? Is it ultimately that you were 1232 01:24:49,120 --> 01:24:51,560 driven a little bit by the problems you were trying to 1233 01:24:51,560 --> 01:24:52,720 solve? I haven't. 1234 01:24:53,280 --> 01:24:56,920 I haven't really thought of it that way, but I think I think 1235 01:24:56,920 --> 01:25:00,440 that's it's correct. Had I, had I been at Stanford, 1236 01:25:00,440 --> 01:25:06,440 say yes, NASA Ames would be the natural point of collaboration 1237 01:25:06,760 --> 01:25:08,760 in research. Who knows? 1238 01:25:08,880 --> 01:25:12,520 I don't. Well, I I say, I don't think I 1239 01:25:12,520 --> 01:25:16,624 would have gone for a one-equation model as Philippe 1240 01:25:16,624 --> 01:25:17,947 Spalart. Yeah. 1241 01:25:17,955 --> 01:25:21,300 And it's interesting, I think now if you look, even with the 1242 01:25:21,308 --> 01:25:24,510 Spalart–Allmaras, there's a lot of modifications to try and add 1243 01:25:24,510 --> 01:25:29,368 some of these non linear terms. So I, I think ultimately people 1244 01:25:29,368 --> 01:25:32,435 come from different directions and you know, I guess you can 1245 01:25:32,443 --> 01:25:34,308 argue which one's correct and which one's not. 1246 01:25:34,316 --> 01:25:37,544 But yeah, it, it's just a curiosity to us externally, 1247 01:25:37,552 --> 01:25:42,272 because if you open up Ansys Fluent or STAR-CCM+ or one of 1248 01:25:42,280 --> 01:25:45,602 these commercial packages, you are almost, you know, this, I 1249 01:25:45,610 --> 01:25:48,660 didn't tell you, but you have radio buttons on which 1250 01:25:48,660 --> 01:25:51,640 turbulence model to pick. Do I pick the k-epsilon? 1251 01:25:51,648 --> 01:25:55,088 Do I pick the k-omega? And in some ways, if you have no 1252 01:25:55,088 --> 01:25:57,721 context of the background, it is in some ways just which one 1253 01:25:57,721 --> 01:26:00,092 do you pick? And that's why people often find 1254 01:26:00,092 --> 01:26:03,035 it so interesting that there's these different options and how 1255 01:26:03,035 --> 01:26:05,960 did they come about and what was their motivation. 1256 01:26:05,968 --> 01:26:10,786 So that ends the question. So maybe to finish off, what if 1257 01:26:10,786 --> 01:26:15,022 you were to now, you know, you're speaking to your 20 year 1258 01:26:15,022 --> 01:26:18,192 old self, what advice would you give having all of what you've 1259 01:26:18,200 --> 01:26:19,840 learned through your amazing career? 1260 01:26:19,848 --> 01:26:22,183 What? What words of wisdom would you 1261 01:26:22,191 --> 01:26:24,754 impart? Gosh, it's hard to imagine a 20 1262 01:26:24,754 --> 01:26:32,520 year old who would sit and listen to somebody closer to 90 1263 01:26:32,520 --> 01:26:36,038 than 80I. I suppose looking back over my 1264 01:26:36,046 --> 01:26:40,656 life, what I can see is that changes have come about. 1265 01:26:40,664 --> 01:26:46,435 Really not of my own driving, but I've somehow been thrust 1266 01:26:46,443 --> 01:26:50,608 upon me. And I think I would say that 1267 01:26:50,616 --> 01:26:56,020 people should be ready to accept change, not be downcast by it 1268 01:26:56,028 --> 01:27:01,680 nor overly ill related, but to be inquisitive of of change. 1269 01:27:01,688 --> 01:27:07,805 And I think making changes constructively has, in a sense, 1270 01:27:07,805 --> 01:27:10,627 reflects my approach to modelling. 1271 01:27:10,635 --> 01:27:15,583 When we'd reached the stage where it was just a matter of 1272 01:27:15,591 --> 01:27:20,020 pushing a button to, as you had alluded to, to choose which 1273 01:27:20,028 --> 01:27:25,076 model is applicable, that's when I think it's time for me to say 1274 01:27:25,076 --> 01:27:28,370 thanks but no thanks, I'll do something else. 1275 01:27:28,378 --> 01:27:33,728 Great words of wisdom and thank you so much for sharing your 1276 01:27:33,736 --> 01:27:37,422 stories and your work. I'm sure I speak for everybody 1277 01:27:37,430 --> 01:27:42,732 to say that the impact of, you know, the k-epsilon, the LRR and 1278 01:27:42,732 --> 01:27:46,980 the wall-function work is today used by hundreds of thousands of 1279 01:27:46,980 --> 01:27:50,440 engineers across the globe to design the things that we and 1280 01:27:50,448 --> 01:27:52,400 you, the listener use it every day. 1281 01:27:52,408 --> 01:27:56,229 So the impact, even though some of that work was 50 years ago, 1282 01:27:56,237 --> 01:27:59,506 it is still impacting today, which is I think an incredible 1283 01:27:59,514 --> 01:28:01,255 testament to the work you've done. 1284 01:28:01,263 --> 01:28:04,265 So thank you for being on today and it was great to speak to 1285 01:28:04,273 --> 01:28:06,076 you. Well, thank you, Neil. 1286 01:28:06,084 --> 01:28:08,240 It's been fun. Bye bye.