Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Tubular cartridge heater wattage

  1. Jun 10, 2017 #1
    I have a machine equipped with 2000mm long tubular cartridge heater (230V, 1600W, 4mm diameter). This tubular heater is sandwiched between the two steel plates (2000mm long) i.e. the heater keeps the steel components hot. My problem is that the temperature of the steel components varies along the heater +/-15degC (nominal T=190degC).

    In order to maintain more uniform temperature along the steel components, I intend to replace the exisitng heater, 2000mm long, with 6 heaters, 325mm long each (there wil be approx. 8mm gap between each nearby 2 heaters). Each 325mm long heater will be powered and controlled separately. But all 6 heaters will be controlled in the way to maintain the same temperature in the steel based on the feedback from 6 thermocouples. The total power (wattage) of all 6x325mm long heaters will be the same as in case of 2000mm heater, i.e. 1600W.
    Please also refer to the attached draft.

    My question is:
    Question 1
    Does it mean that I would need to basically divide the power of 2000mm long heater, i.e. 1600W, by 6? What gives me approx. 270W per heater.
    (If yes, I would then add perhaps 20% of power to each heater for some headroom, i.e. 270W+20%= 324W).
    I have some doubt in regards to this solution because 6 heaters will be spread on the length of 2000mm and I am not sure whether simple dividing 1600W by 6 is adquate here.
    Question 2
    Do I maintain the in-line control on a single 325mm long heater temperature basically through varying the input voltage delivered to it?
    If yes, my requirement to the heater supplier would be the condition that 325mm long heater must be capable to be operational at varying input voltage, say 180V-280V (with 230V nominal).

    Please help.

    Regards Untitled1.png Untitled1.png
  2. jcsd
  3. Jun 11, 2017 #2
    The text indicates a 2000mm long heater is currently used, but the drawing shows it as 1600mm.

    Where is the present thermocouple located?
    Where will the (6) planned thermocouples be located?
    These steel plates that are heated - are they parts in the machine?

    The idea of dividing total power into the number of heaters is valid. However, cold spots occur where heaters butt up against one another in the 'planned' drawing, partially due to the spacing between them, and partially because each heater element has a 4 to 8 mm dead spot at each end for the electrical terminations. Is it possible to stagger the heaters so they overlap?
    While it is possible to vary voltage by using a phase-fired power controller, the practical drawbacks of electrical noise generation and cost usually result in ceding it's use to applications that absolutely require it. More often, heater power control is time proportioned by adjusting the length of on-time proportional to total cycle time. For instance, if cycle time is 8 seconds, then the heater is turned on for 4 seconds to obtain 50% output, 2 seconds for 25% output, 8 seconds (always on) for 100%, and so on.
    Heaters are rated at full operating voltage (i.e., 230V), and not for a variable range (i.e., 180V-280V). If a heater is expected to withstand operation at 280V, it is rated for 280V.
  4. Jun 11, 2017 #3
    Thank you for your reply.
    The actual length of currently used heater is 2000mm (the length shown on the drawing is incorrect).
    Presently 1 thermocouple is located close to one of the far ends of the steel plate (outside of the plate).
    (The temperature variation along the steel plates was monitored with 8 thermocouples (K type) fixed at equal distances.)
    Each of the planned thermocouples will be fixed to the outside wall of the steel plate ''in the centre" of each 325mm long heater.
    The steel plates are the components of the machine. Unfortunately, it would be very challenging to overlap the heaters to avoid the cold spots.
    These plates have hundreds of holes and grooved channels (where the polymer flows) so there is no space for other than linear array of short (325mm long) heaters.
    Thank you also for your answers in regards to the temperature control.
  5. Jun 11, 2017 #4

    jim hardy

    User Avatar
    Science Advisor
    Gold Member

  6. Jun 11, 2017 #5
    I'm trying to visualize what this installation looks like in cross-section. Do the plates tangentially touch the outer heater surface, or are they machined so the heater is in good contact with them along the lines of this illustration?

    If the latter, have you tried using a heat transfer compound between heater and plate? For this temperature range, very finely powered graphite and aluminum in a silicone or mineral oil base has been successfully employed, and is commercially available. If temperature variation is caused by uneven contact between heater and plates this should bring it under control.
  7. Jun 11, 2017 #6

    jim hardy

    User Avatar
    Science Advisor
    Gold Member

    Why is that ? Does cold process fluid enter at one end ? Might a little bit of recirculation preheat incoming fluid ?

    It's time lag that makes temperature control difficult. Thermal capacity and transport time from heater to sensor cause overshoot and hunting. The closer you can place your sensors to the heaters the better.
  8. Jun 11, 2017 #7
    @jim hardy
    Thank you for the link to the controllers brochure. It is very useful.
    The molten polymer (180degC) which folws through the steel plates has its inlet at one of the plate ends. The polymer flow channels are not 'symmetrical' along the steel plates length (2000mm) and it is supposedly one of the thermal issues here.
    The two steel plates are part of a larger assembly and more uniform temperature is required also in other nearby parts of the assembly. Ideally the 6 thermocouples would be embeded inside the heater. However, there are various factors which affect 'locally' uniformity of the temperature and placing the thermocouples 'too close' to the heater may mask the thermal problems residing in the assembly. Although indeed, possible decay in the heater 'actual response' to the thermocouple feedback may not be optimum either.

    Thank you for your question.
    Unfortunately, the channel which accommodates the heater is of square shape and is too deep, please refer to the attached draft on left side, therefore is not ideal for the round heater.
    The steel parts are of old design but for a moment I need to go ahead with them because they are too expensive.
    I considered two actions in order to temporarily fix the problem of square channel and round heater. Please refer to the draft on right side.
    I think about making the channel slightly deeper with a round mill (4mm dia), so at least half of the channel (bottom part) would be in a good contact with the heater and then I would add a mild steel strip over the heater. The compound would be added there. However, the problem is that the two steel plates needs to be firmly tighted (there is 1mm aluminium gasket between them) two avoid any polymer or hot compressed air (which also flows through hundreds of channels in these plates) leakage. Thus the strip inside the channel would need to be only in a slight touch with the heater.
    Untitled1.jpg Untitled1.jpg
  9. Jun 11, 2017 #8
    Do you have issues with limited heater lifespan, too? 1600W in a 2000mm x 4mm tube is about 41 watts/square inch of watt density. This yields higher than desirable sheath temperatures wherever heater-to-heated object contact is poor.

    I'd try rounding the bottom of the channel with a 4mm ball mill as you've described. Rather than a flat strip on top, however, make it appropriately deeper, and round it with the same 4mm ball mill so it also is in good thermal contact. Do this before experimenting with multiple elements and heat zones - basic geometry improvements will go a long way to even out thermal distribution within the plates, and perhaps enough that multiple zones will no longer be necessary.
  10. Jun 11, 2017 #9

    jim hardy

    User Avatar
    Science Advisor
    Gold Member

    Indeed a temperature measurement as you describe that's ahead of the process's thermal inertia is de-facto feedforward.
    I see you have understanding of your process..

    I guess i dont quite understand why the structure is steel rather than aluminum for best heat transfer. Do you need the strength ?

    old jim
  11. Jun 11, 2017 #10
    When @AligatorAmy said "These plates have hundreds of holes and grooved channels (where the polymer flows)" and "The molten polymer (180degC) which flows through the steel plates has its inlet at one of the plate ends. The polymer flow channels are not 'symmetrical' along the steel plates length (2000mm) and it is supposedly one of the thermal issues here." it put me in mind of injection molding.

    Injection molding is a completely different critter than sheet thermoforming (which I've had experience with) but from what I've picked up by talking to folks in that field, and in an article I've just read, strength factors into it, as do coefficient of expansion, and wear rate.
  12. Jun 11, 2017 #11
    Is it possible the tool builder originally designed it using a square or rectangular heater? I've never used them (in fact, I didn't know they existed until now), but several firms manufacture such heaters. Here are links to two of them.

    4.2mm x 5.3mm isn't a standard size, and 3/16" (4.76mm) is the smallest I've seen so far (there may be issues fabricating cartridge elements under this size), but it's worth discussing the possibility of manufacturing a custom rectangular heater with your supplier.
  13. Jun 12, 2017 #12


    User Avatar
    Science Advisor

    Or how about filling/packing that void around the heater with ceramic powder as a thermal transfer medium? Possibly without even having to disassemble it.

    P.S. Rather than paying for ceramic powder, you could even use sand as used for sand blasting.
  14. Jun 12, 2017 #13
    Not a bad idea, so long as the powder is very fine (like flour) and tightly packed. Usually magnesium oxide is used due to high thermal conductivity for an electrical non-conductor (45 to 60 W/mK).

    If the blasting 'sand' is aluminum oxide and has been pulverized and spent (not fresh - the grain size is too large pack effectively) it'll work nearly as well (~35 W/mK thermal conductivity), but good old-fashioned silica sand won't (~2 W/mK thermal conductivity).
  15. Jun 12, 2017 #14
    The lifetime of the heaters does not seem to be short. Although I would need to have round finish of the heater channel first and introduce it into the production, for comparison.
    I will definitely try the option with re-mill of the channel bottom (4mm dia ball mill) along with application of the rounded top strip sitting on the heater.

    @Asymptotic, jim hardy
    The steel parts are the components of the melt blowing equipment. The plates and other components design is outdated and 'not perfect' and the unsymmetry of the polymer flow channels along the plates is only one example. For this moment I need to go ahead with current design because re-design and manufacturing of new components would automatically enforce change of others...

    @Asymptotic, Tom G
    Application of the compound is defintely worth consideration. Thank you for the examples of the materials that you have suugested.
  16. Jun 22, 2017 #15
    I have decided to go ahead with the solution involving split of 2000mm long heater into the 6 separate sections (6 heaters, 320mm long each).
    I am looking for the PID temperature controllers for these heaters. Preferably 1 controller which could accomodate at least 6 inputs (from TCKs). I have came across Eurotherm company and multiple analog input T controller (with 8 in/out):


    Do you recommend other providers/makes?

    P.S. The input from all 6 TCKs would need to be simultaneously displayed on a screen of such a controller to allow the operator easy access to cureent temperatures status.
  17. Jun 22, 2017 #16
    A Eurotherm 2704 handles up to three control loops each, so you'd need two of them.
    It would probably be cheaper to use six single loop controllers. Among the least expensive are 1/32 DIN sized model SL4824 for $90/ea marketed by AutomationDirect.
  18. Aug 24, 2017 #17
    I would also look into www.watlow.com controllers. Asymptotic is probably right, 6 individual controllers may be cheaper and easier to setup. If you want to go the fancy route, Watlow also makes an F4T touch screen controller.
  19. Aug 27, 2017 #18


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member

    Ok so it's 15 degrees hotter at one end compared to the other. That works out at 7.5C/m. If you replace it with 6 heaters each 0.3m (and each temperature controlled) I would expect the gradient along each strip to be reduced to around 7.5*0.3 = 2.25C. That's a crude calculation. Could be less as the hot end of one section should raise the cold end of the adjacent section.

    Re post #7: Copper strip instead of mild steel?
  20. Aug 27, 2017 #19
    He likely means the sheath temperature varies 15C at any given point along the length of the element, not necessarily a relatively uniform gradient. The coiled wire inside a tubular element is vibrated while filling with MgO, and then roll reduced, so the coil is rarely centered on the heater and not always uniformly stretched (due to gravity and vibration during the fill process).

    6 smaller heaters, set to the same setpoint temperature, may not fix this issue. There will be finer control over individual zones, however, which should still reduce the variability that OP is looking for. Down to 2.25C? Highly doubtful, and likely unmeasureable.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted