<html>
  <head>
    <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <p>One comment - my dissertation below is specifically about
      non-ebullient immersion cooling. As Jim Lux pointed out in a later
      e-mail, in ebullient cooling, some kind of surface feature to
      promote nucleation could be beneficial. Ebbulient cooling is a
      whole different beast from normal (non-ebullient) immersive
      cooling, since in that case you have changes of state and gas
      bubbles flowing through a liquid. <br>
    </p>
    <p>However, in all of the live and video demonstrations I've seen of
      Novec, the processors were completely bare, bubbles were forming
      at a pretty rapid rate, so again I think creating some sort of
      heat sink for this would add cost with no significant benefit. <br>
    </p>
    <pre class="moz-signature" cols="72">Prentice Bisbal
Lead Software Engineer
Princeton Plasma Physics Laboratory
<a class="moz-txt-link-freetext" href="http://www.pppl.gov">http://www.pppl.gov</a></pre>
    <div class="moz-cite-prefix">On 11/08/2018 10:40 AM, Prentice Bisbal
      wrote:<br>
    </div>
    <blockquote type="cite"
      cite="mid:f1f082d1-e665-9a0c-2349-70ca229b6297@pppl.gov">
      <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
      <p>Heat fins are used to increase the surface area used for heat
        transfer, since the rate of energy transfer by conduction is
        directly proportional the surface area. Heat fins are needed
        when air is involved because air has such a low thermal
        conductivity. <br>
      </p>
      <p>Thermal conductivity of liquids are much high, so heat fins
        aren't as necessary. For example, I've read that water can
        transfer heat orders of magnitude better than air, so using
        water to remove hear from a processor would need orders of
        magnitude less surface area for the same energy transfer rate. <br>
      </p>
      <p>Also, liquids have higher viscosities than gases, so we have to
        worry about 'boundary layers'. A boundary layer is area where
        the edge flowing fluid is in contact with a solid. The friction
        between the liquid and the solid slows down the fluid near the
        solid. This affects both gases and liquids, but since liquids
        have higher viscosities, the effect is more noticeable. <br>
      </p>
      <p>Think about a car's radiator - the air side has all the fins on
        it, and the liquid side has smooth pipe walls. <br>
      </p>
      <p><a class="moz-txt-link-freetext"
          href="https://en.wikipedia.org/wiki/Boundary_layer"
          moz-do-not-send="true">https://en.wikipedia.org/wiki/Boundary_layer</a></p>
      <p>Convection heat transfer is an equally important mode of heat
        transfer in fluids, and in the boundary layer, where the liquids
        aren't moving as fast, heat transfer isn't as good, so you need
        to keep your boundary layer from becoming too thick. <br>
      </p>
      <p>Since fluids have much higher thermal conductivities, and
        boundary layer effects are more of a concern, I actually think a
        smooth heat transfer surface would be better in these immersion
        cooling cases. I'm sure smaller,  more spaced out fins would
        probably help heat transfer without  creating too much of a
        boundary layer, but making those heat sinks adds cost for
        increased performance in a situation where it probably isn't
        needed. <br>
      </p>
      <p>Now direct-contact cooling systems like Asetek products  do
        have fins on the liquid side, if I remember correctly, but that
        in those systems, there are pumps to provide forced convection.
        In immersion cooling, you are relying on natural convection, so
        there isn't as much driving force to overcome viscosity/boundary
        layer effects to force the liquid through the heat fins. <br>
      </p>
      <p>That's my thoughts, anyway. <br>
      </p>
      <pre class="moz-signature" cols="72">Prentice </pre>
      <div class="moz-cite-prefix">On 11/07/2018 04:12 AM, John Hearns
        via Beowulf wrote:<br>
      </div>
      <blockquote type="cite"
cite="mid:CAPqNE2XO_99NQ-3GP1ZixwfFoFzmFUGNMKErDHdRxHWB8e=AzQ@mail.gmail.com">
        <meta http-equiv="content-type" content="text/html;
          charset=utf-8">
        <div dir="ltr">
          <div dir="ltr">
            <div>Thinking about liquid cooling , and the ebuillient
              cooling, the main sources of heat on our current
              architecture servers are the CPU package and the voltage
              regulators. Then the DIMMs.</div>
            <div>Concentrating on the CPU die package, it is engineered
              with a flat metal surface which is intended to have a
              thermal paste to transfer heat across to a flat metal
              heatsink.</div>
            <div>Those heatsinks are finned to have air blown across
              them to transport the heat away.</div>
            <div><br>
            </div>
            <div>In liquid immersion should we be looking at having a
              spiky surface on the CPU die packages and the voltage
              regulators?</div>
            <div>Maybe we should spray the entire board with a
              'flocking'' compound and give it a matt finish!</div>
            <div>I am being semi-serious. I guess a lot of CFD
              simulation  done regarding air cooling with fins.</div>
            <div>How much work has gone into pointy surfaces on the die
              package, which would increase contact area of course and
              also act as nucleation points for bubbles?</div>
            <div><br>
            </div>
            <div>One interesting experiment to do - assuming the flat
              areas of the CPU in an immersive system do not have (non
              thermal paste) heatsinks bolted on:</div>
            <div>take two systems and roughen up the die package
              surfacewith sandpaper on one. Compare temperatures.</div>
            <div><br>
            </div>
            <div>ps. I can't resist adding this. Sorry Stu .   <a
                href="https://www.youtube.com/watch?v=kHnifVTSFEo"
                moz-do-not-send="true">https://www.youtube.com/watch?v=kHnifVTSFEo</a></div>
            <div>I guess Kenneth Williams is a typical vendor Site
              Engineer.</div>
            <div>pps. the actress in the redress had her career ruined
              by this film - she ver got a serious role again after
              perfectly being typecast.</div>
            <div><br>
            </div>
            <div><br>
            </div>
            <div><br>
            </div>
            <div><br>
            </div>
            <div><br>
            </div>
            <div><br>
            </div>
          </div>
        </div>
        <br>
        <div class="gmail_quote">
          <div dir="ltr">On Tue, 6 Nov 2018 at 22:57, Prentice Bisbal
            via Beowulf <<a href="mailto:beowulf@beowulf.org"
              moz-do-not-send="true">beowulf@beowulf.org</a>> wrote:<br>
          </div>
          <blockquote class="gmail_quote" style="margin:0 0 0
            .8ex;border-left:1px #ccc solid;padding-left:1ex">
            <div text="#000000" bgcolor="#FFFFFF">
              <div class="m_4669408001860479086moz-cite-prefix">On
                11/06/2018 02:03 PM, Lux, Jim (337K) wrote:<br>
              </div>
              <blockquote type="cite">
                <div class="m_4669408001860479086WordSection1">
                  <p class="MsoNormal"><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">True
                      enough.</span></p>
                  <p class="MsoNormal"><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d">Ebullient
                      cooling does have some challenges – you can form
                      vapor films, which are good insulators, but if you
                      get the system working right, nothing beats phase
                      changes for a heat transfer.  </span></p>
                </div>
              </blockquote>
              If I recall what I learned in my Transport Phenomena
              classes in engineering school, you need a reasonably high
              temperature difference to get a stable film like that. For
              that to happen, radiant heat transfer needs to be the
              dominant heat transfer mechanism, in the range of
              operation we are talking about, the temperature difference
              isn't that great, and conduction is still the dominant
              form of heat transfer. <br>
              <br>
              Here's an example of what 3M Novec ebullient cooling looks
              like. It doesn't look like it's anywhere near the film
              boiling regime: <br>
              <br>
              <a class="m_4669408001860479086moz-txt-link-freetext"
                href="https://www.youtube.com/watch?v=CIbnl3Pj15w"
                target="_blank" moz-do-not-send="true">https://www.youtube.com/watch?v=CIbnl3Pj15w</a><br>
              <br>
              --<br>
              Prentice<br>
              <br>
              <blockquote type="cite">
                <div class="m_4669408001860479086WordSection1"><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"></span>
                  <p class="MsoNormal"><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"> </span></p>
                  <span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"></span>
                  <p class="MsoNormal"><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:#1f497d"> </span></p>
                  <div>
                    <div style="border:none;border-top:solid #e1e1e1
                      1.0pt;padding:3.0pt 0in 0in 0in">
                      <p class="MsoNormal"><b><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext">From:</span></b><span
style="font-size:11.0pt;font-family:"Calibri",sans-serif;color:windowtext">
                          Beowulf [<a
                            class="m_4669408001860479086moz-txt-link-freetext"
                            href="mailto:beowulf-bounces@beowulf.org"
                            target="_blank" moz-do-not-send="true">mailto:beowulf-bounces@beowulf.org</a>]
                          <b>On Behalf Of </b>Prentice Bisbal via
                          Beowulf<br>
                          <b>Sent:</b> Tuesday, November 06, 2018 8:17
                          AM<br>
                          <b>To:</b> <a
                            class="m_4669408001860479086moz-txt-link-abbreviated"
                            href="mailto:beowulf@beowulf.org"
                            target="_blank" moz-do-not-send="true">beowulf@beowulf.org</a><br>
                          <b>Subject:</b> Re: [Beowulf] More about those
                          underwater data centers</span></p>
                    </div>
                  </div>
                  <p class="MsoNormal"> </p>
                  <blockquote
                    style="margin-top:5.0pt;margin-bottom:5.0pt">
                    <p class="MsoNormal">. And serviceability is
                      challenging. You need to pull the "wet" boards
                      out, or you need to connect and disconnect fluid
                      connectors, etc.  If you're in an environment
                      where you can manage that (or are forced into it
                      by necessity), then you can do it.</p>
                  </blockquote>
                  <p>I think everyone on this list already knows I'm no
                    fan of mineral oil immersion (It just seems to messy
                    to me. Sorry, Stu), but immersion cooling with other
                    liquids, such as 3M Novec engineered fluid addresses
                    a lot of your concerns. It as a low boiling point,
                    not much above room temperature, and it was
                    originally meant to be an electronic parts cleaner
                    (according to a 3M rep at the 3M booth at SC a few
                    years ago, so if you pull a component out of it, it
                    dries very quickly and should be immaculately clean.
                  </p>
                  <p>The low boiling point is an excellent feature for
                    heat transfer, too, since it boils from the heat of
                    the processor (ebullient cooling). This change of
                    state absorbs a lot of energy, making it very
                    effective at transferring heat away from the
                    processor. The vapor can then rise and condense on a
                    heat exchanger with a chilled water heat exchanger,
                    where it again transfers a lot of heat through a
                    change of state. </p>
                  <pre>Prentice </pre>
                  <div>
                    <p class="MsoNormal">On 11/05/2018 06:30 PM, Stu
                      Midgley wrote:</p>
                  </div>
                  <blockquote
                    style="margin-top:5.0pt;margin-bottom:5.0pt">
                    <div>
                      <div>
                        <p class="MsoNormal">I refute both these claims.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal">You DO want to run your
                          boards immersed in coolant.  It works
                          wonderfully well, is easy to live with,
                          servicing is easy... and saves you almost 1/2
                          your power bill.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal">People are scared of
                          immersion cooling, but it isn't that difficult
                          to live with.  Some things are harder but
                          other things are way easier.  In total, it
                          balances out.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal">Also, given the greater
                          reliability of components you get, you do less
                          servicing.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal">If you haven't lived with
                          it, you really have no idea what you are
                          missing.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal">Serviceability is NOT
                          challenging.</p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <div>
                        <p class="MsoNormal"> </p>
                      </div>
                      <p class="MsoNormal"> </p>
                      <div>
                        <blockquote style="border:none;border-left:solid
                          #cccccc 1.0pt;padding:0in 0in 0in
                          6.0pt;margin-left:4.8pt;margin-right:0in">
                          <p class="MsoNormal">You really do NOT want to
                            run boards immersed in coolant - yeah,
                            there's folks doing it at HPC scale<br>
                            <br>
                            Whatever the coolant, it leaks, it oozes, it
                            gets places you don't want it to go. And
                            serviceability is challenging. You need to
                            pull the "wet" boards out, or you need to
                            connect and disconnect fluid connectors,
                            etc.  If you're in an environment where you
                            can manage that (or are forced into it by
                            necessity), then you can do it.</p>
                        </blockquote>
                      </div>
                      <p class="MsoNormal">-- </p>
                      <div>
                        <div>
                          <p class="MsoNormal">Dr Stuart Midgley<br>
                            <a href="mailto:sdm900@gmail.com"
                              target="_blank" moz-do-not-send="true">sdm900@gmail.com</a></p>
                        </div>
                      </div>
                    </div>
                    <p class="MsoNormal"><br>
                      <br>
                      <br>
                    </p>
                    <pre>_______________________________________________</pre>
                    <pre>Beowulf mailing list, <a href="mailto:Beowulf@beowulf.org" target="_blank" moz-do-not-send="true">Beowulf@beowulf.org</a> sponsored by Penguin Computing</pre>
                    <pre>To change your subscription (digest mode or unsubscribe) visit <a href="http://www.beowulf.org/mailman/listinfo/beowulf" target="_blank" moz-do-not-send="true">http://www.beowulf.org/mailman/listinfo/beowulf</a></pre>
                  </blockquote>
                  <p class="MsoNormal"> </p>
                </div>
              </blockquote>
              <br>
            </div>
            _______________________________________________<br>
            Beowulf mailing list, <a href="mailto:Beowulf@beowulf.org"
              target="_blank" moz-do-not-send="true">Beowulf@beowulf.org</a>
            sponsored by Penguin Computing<br>
            To change your subscription (digest mode or unsubscribe)
            visit <a
              href="http://www.beowulf.org/mailman/listinfo/beowulf"
              rel="noreferrer" target="_blank" moz-do-not-send="true">http://www.beowulf.org/mailman/listinfo/beowulf</a><br>
          </blockquote>
        </div>
        <br>
        <fieldset class="mimeAttachmentHeader"></fieldset>
        <br>
        <pre wrap="">_______________________________________________
Beowulf mailing list, <a class="moz-txt-link-abbreviated" href="mailto:Beowulf@beowulf.org" moz-do-not-send="true">Beowulf@beowulf.org</a> sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit <a class="moz-txt-link-freetext" href="http://www.beowulf.org/mailman/listinfo/beowulf" moz-do-not-send="true">http://www.beowulf.org/mailman/listinfo/beowulf</a>
</pre>
      </blockquote>
      <br>
    </blockquote>
    <br>
  </body>
</html>