Measuring junk sailing performance

<< First  < Prev   1   2   3   4   5   ...   Next >  Last >> 
  • 08 Oct 2020 15:49
    Reply # 9291787 on 8083678
    Anonymous member (Administrator)
    Anonymous wrote:
    Anonymous wrote:

    Correct me if I'm wrong, but I don't think Robert is applying any downwash correction to the data. I think he is just using the raw recorded data.

    The correction for Amiina's length and sail area and displacement does suggest that the formula we are using does not work across the range of boats tested, so I have decided to just use the square root of sailing length = (2 x waterline length + overall length)/3 for future comparisons, and ignore sail area displacement ratio effects, which I now think are second order effects.

    Hi Alan--

    Correct I use the raw data after transformation from apparent to true.

    I do search for and remove data lines/sentences that are duplicates, i.e. exact same date-time stamp (and column values) and lines that are missing column values. This cull only removes a fraction-of-1-percent of the original data.

    Re: sailing length. Do I understand that the factor you'll apply is this line in your spreadsheet?

                                         weaverB, calisto,miranda,poppy,amiina

    "Factor to Sqrt LWL of 5     1.13    1.04    1.11    0.96    1.17"

    Rather than this line:

    "Factors combined             1.05    1.14    1.29    0.94    1.48"

     robert self

    Yes, that's right. The combined factors included a correction for SA/Disp ratio, but it was excessively weighted and had too much effect, and is quite debatable whether it is valid anyway, as upwind stability is more significant that SA/D, so we now just use the length correction.
  • 08 Oct 2020 15:36
    Reply # 9291763 on 4913961
    Anonymous member (Administrator)

    Finally ... we have the polar diagram results for all the boats tested to date. I apologise for taking so long to produce these, but since 2019 Anthony Cook has been working to "automate" the analysis of the results, and the creation of the polar diagrams, which I used to produce by hand.

    Attached is the 7 boat multi boat comparison sheet, with polar diagrams at 2 knot windspeed steps from 6 knots up to 16 knots, displayed for port and starboard tacks as though the wind is blowing down from the top of the chart. Some boats did not have any data for some wind angles at some speeds and tacks, so there are some "missing" bits of the curves. The speeds have been adjusted for sailing length to 25ft.

    The curves have been smoothed by finding the "high points" in the filtered data and interpolating between them linearly. You can see how this is done in the individual sheets for Gung Ho and Weaverbird WS.

    I have also calculated the VMG for each boat, wind speed, and wind angle between 20 and 85 degrees (true), and extracted the maximum values for port and starboard tacks, and plotted them against wind speed so you can see exactly how the windward performance compares at different wind speeds. I have excluded VMG values below 2 knots because where they occur, they are caused by a lack of data in the upwind sector.

    By taking the average VMG between 8 knots and 16 knots, and across port and starboard, we can get to a single figure for windward performance, which could be called "Average Max VMG" This is obviously a great simplification, but is interesting and has some validity as a quick comparison.

    Combined P & S

    Average of VMGs for

    8-16 kts windspeed

    as % of best
    Callisto Bermudan 3.64 100.0
    Miranda Aerojunk 2.24 61.5
    Weaverbird HB Hinged battens 2.86 78.6
    Poppy Split junk 2.91 79.9
    Amiina Split junk 2.93 80.6
    Weaverbird WS Wingsail 2.80 77.0
    Gung Ho AB Flat schooner 2.74 75.4

    You could conclude from this that in normal sailing conditions, and taking account of both tacks, most of the junks make ground to windward at between 75% and 80% of the speed of the bermudan rig. Of course, there are lots of caveats to this, the main one being the lack of data in some cases.

    Also, you need to be aware that all the boats except Gung Ho have been corrected for the downwash effect on the wind sensor mounted aft.

    The AB after Gung Ho indicates that this polar is made up from two sets of data. A data is with the wind sensor Aft, used from 90 to 180 degrees, and B data is with the wind sensor in the Bow, used for the part of the curve from 0 to 90 degrees.

    What is also interesting about these polars is that they show up quite clearly the differences in performance between port and starboard tacks for those boats that have the mast on one side of the sail. These boats are always worse to windward when the mast is to leeward, which is no surprise, but it is interesting to see it quantified.

    I would very much like to re-test all the boats except Gung Ho with the wind sensor in the bow so we get more data to compare, and more accurate data to windward. I also want to test a cambered panel sail, or two. That's for next year.

    5 files
    Last modified: 08 Oct 2020 15:46 | Anonymous member (Administrator)
  • 30 Oct 2019 13:24
    Reply # 8083711 on 4913961
    Anonymous member (Administrator)

    Alan,
    it appears that you are right that moving the wind sensors well forward of the sail will correct things. I suggested that the sensor should sit a whole chord’s distance from the sail, but maybe that was a bit over the top.

    Since the algorithms seem to do their jobs correctly, it must have been the sensors, sitting in the ‘bent wind’ aft of the sail, which fooled us to believe that the boats were sailing so (impossibly) close to the wind.

    Arne


  • 30 Oct 2019 13:01
    Reply # 8083678 on 8082413
    Anonymous wrote:

    Correct me if I'm wrong, but I don't think Robert is applying any downwash correction to the data. I think he is just using the raw recorded data.

    The correction for Amiina's length and sail area and displacement does suggest that the formula we are using does not work across the range of boats tested, so I have decided to just use the square root of sailing length = (2 x waterline length + overall length)/3 for future comparisons, and ignore sail area displacement ratio effects, which I now think are second order effects.

    Hi Alan--

    Correct I use the raw data after transformation from apparent to true.

    I do search for and remove data lines/sentences that are duplicates, i.e. exact same date-time stamp (and column values) and lines that are missing column values. This cull only removes a fraction-of-1-percent of the original data.

    Re: sailing length. Do I understand that the factor you'll apply is this line in your spreadsheet?

                                         weaverB, calisto,miranda,poppy,amiina

    "Factor to Sqrt LWL of 5     1.13    1.04    1.11    0.96    1.17"

    Rather than this line:

    "Factors combined             1.05    1.14    1.29    0.94    1.48"

     robert self

    Last modified: 30 Oct 2019 13:13 | Anonymous member
  • 29 Oct 2019 16:58
    Reply # 8082413 on 4913961
    Anonymous member (Administrator)

    Correct me if I'm wrong, but I don't think Robert is applying any downwash correction to the data. I think he is just using the raw recorded data.

    Arne, although the correction is arbitrary, it is considered, and is based on the idea that the optimum VMG for an averagely good windward boat should/will be around 45 degrees true wind direction, and about 30 degrees apparent. On the basis of the original data we had last year, it looked as though a 10 degree correction at 35 degrees would achieve the desired result. The question then was how to taper that off, and I figured the downwash effect should be minimal at 70 degrees TWD. So we made it taper to zero at 70 degrees.

    I would happily admit this is a far from perfect solution, but it does allow some comparisons to be made between the boats we have tested, provided we use the same correction for all of them.

    The next step, I think, for next year, is to arrange to mount the wind sensor well forward of the sail, but I think a full chord length will be hard to achieve. I also think the upwash effect will not extend more than half a chord length forward of the sail, and probably less. I am thinking we should try to get it 10 to 12 ft forward of the stem, if we can.

    By re-testing the boats we have already tested, we can assess whether the downwash varies from boat to boat, and hopefully get some better upwind data. I am not concerned about the downwind data.

    The correction for Amiina's length and sail area and displacement does suggest that the formula we are using does not work across the range of boats tested, so I have decided to just use the square root of sailing length = (2 x waterline length + overall length)/3 for future comparisons, and ignore sail area displacement ratio effects, which I now think are second order effects.

  • 14 Oct 2019 18:32
    Reply # 7959583 on 4913961

    Well, before the output from the instrumentation is throughly trashed I'll publish my summary graphs and opinions.

    The new boats added this year were Weaverbird fitted with a new wing sail JR (same sail area as the 2018 single-ply JR) and Amiina with a split JR.

    Weaverbird comparisons;

    Upwind the wing JR and single-ply, hinged, cambered JR are nearly equal.

    On a beam reach I'd say a slight advantage goes to the single-ply although there is substantial overlap.

    Down wind the advantage goes to the wing JR.

    In the down wind plot note the small print in the individual panel titles. My attempt to uncover bias in the frequency distributions due to inadequate sample sizes for high or low wind speeds. The ideal is an equal number of measurements in all combinations of TWA and TWS. But there is no way to fully control the wind speed and course sailed on the day data is collected. The idea is to make the viewer aware. Up to them to disregard, in part or in whole, the graph. For example Amiina is missing data from 6-9 knt wind speeds. Only showing data for 10,11,12 knt wind speed. Higher windspeed, boat sail faster. So there is an inherit upward bias to the Amiina graph, out of the control of the data takers, when the viewer compares Amiina boat speeds to boats which are not misssing wind speeds in this category. Perhaps, if the 6-9 knt data had been collected the Amiina frequency distribution would extend down into the 2 knt boat speed region.

    Amiina comments;

    I found Amiina a surprising overall performer. Alan's scalar says it all (see spec spreasheet below). The higher the scalar the slower the expected boat speed. The scalar is an imperfect handicapper or rating designed to normalize boats of unequal length, weight, sail area, etc.

    That said, Amiina has the shortest waterline hence lowest maximum speed. Moderate D/L of 252 hence is not a light-displacement planer like a X-99. Low-ish ballast ratio of 32% not indicative of standing up to the wind when beating like a folkboat with 50% ratio. Lowest SA/D of 14. Under-powered relative to the other sailboats. By these numbers Amiina's "as measured" boat speeds should be measureably less than the other designs. But they are not. Amiina performs on level upwind and abeam. Down wind speeds are potentially biased high.

    Applying the scaling factors (red histograms), intended to achieve an apples-to-apples comparison, Amiina is a visible outperformer. Unless there has been some radical changes to Amiina compared to the Splinter 21 specs on the sailboatdata page. Perhaps something is going on here(?) i.e. the split JR.

    view and download link:

    https://drive.google.com/drive/folders/1b3zneYJrA8ATxeI8bsgFa3Q2025VAFEy

    robert self


    4 files
    Last modified: 14 Oct 2019 22:06 | Anonymous member
  • 14 Oct 2019 16:39
    Reply # 7959405 on 4913961
    Anonymous member (Administrator)

    Robert, I have no problem with the orientation of the diagram. Whether the wind comes in from top or from right is just the same for me. It is the shown results, which I think is un-achievable for a Splinter (with any rig), I have problems with.

    I have eye-balled quite a number of polar diagrams to learn about the windward performance of different boats, and their VMGs to windward seem to culminate at between 40 and 50 degrees from the (true) wind. 

    Even a 6mR would struggle with 30 degrees, not to mention having best VMG at that angle.

    Arne

    Last modified: 16 Oct 2019 13:18 | Anonymous member (Administrator)
  • 14 Oct 2019 16:18
    Reply # 7959390 on 7958948
    Anonymous wrote:

    Robert,
    that polar diagram of Weaverbird makes sense. It shows that best VMG is reached at 40-50° from the wind, and is dropping at 30° from the wind. I even understand the green VMG-dots, although I don’t see a need for them as they are just the horizontal component of the STW curve (vertical component on the Contessa’s diagram).

    However, several of Amiina’s polar diagrams indicate that she is at her best when pointing only 30° from the wind.
    Amiina has shown before that she sails smartly to windward, but there is nothing supernatural about her close-windedness  -  which the polar diagram below appears to indicate.
    This is why I suspected that there was something wrong with the algorithm.

    Arne


    Arne--OK.

    I did not realize this orientation would be so confusing. Between now and next year when (hopefully) data from new boats becomes available I'll look for a matlab prog which defaults to the usual sailing orientation--0 degrees up and clockwise rotation.

    Last modified: 14 Oct 2019 16:22 | Anonymous member
  • 14 Oct 2019 14:12
    Reply # 7959186 on 7959022
    Anonymous member (Administrator)
    Alan wrote:

    The problem here is that. as we discovered last year, the wind sensor mounted on the transom is affected by downwash from the sail when sailing close hauled.

    Last year we developed the idea of correcting the TRUE wind values between 0 and 70 degrees so they are reduced by 10 degrees at 35 degrees off the wind, and the correction tapers off to zero at 0 degrees and at 70 degrees.

    This is a completely arbitrary correction to take care of the perceived affect of downwash on the results, and just adjusts the clearly incorrect upwind performance values to something more reasonable.



    During my life as an electronic engineer, regularly checking the integrity of critical navigational aids on (mostly) Stavanger Airport, I and my collegues learned to be aware of any error sources.
    We used to say -
    “Garbage in results in garbage out”

    This is why I became sceptical to those polar diagrams. To just add a suitable (..‘arbitrary’  -  now I learned a new word...) correction factor in the ‘think-of-a-number’ class, is no good. The upwash and downwash error will no doubt vary with the sailing angle, as the sensor’s vane will alter position with respect to the sail.

    It therefore seems to me that the found results have little value, so one is back to square one.

    My guess is that the wind sensor needs to be moved at least one sail chord’s length away from the sail (I suggest in the bow) to minimise the error caused by the upwash.
    If I remember correctly, the speed (STW) sensor was calibrated with a GPS in slack waters. In that case, we can at least trust that sensor.

    Good luck!

    Arne

    PS:
    With the wind sensor on a long stick, set forward of the bow, the sensor will of course be blocked when running straight before. Still, I would rather sacrifice measurements downwind to achieve good data from a broad reach and up to fully close-hauled.



    Last modified: 14 Oct 2019 16:34 | Anonymous member (Administrator)
  • 14 Oct 2019 12:05
    Reply # 7959065 on 7959041
    Anonymous member (Administrator)
    Anonymous wrote:

    I agree, Arne, there’s something wrong somewhere. My observations as a seat of the pants sailor are that Weaverbird points a little closer, and has more power through the seas, with the wing sail than with JR; and that Weaverbird points a little higher than Amiina but is no faster. I’m not seeing that reflected in the diagrams. 

    Alan, I found that there was a lot of upwash with the wind sensor on the bow, so that doesn’t help.   


    I guess it would need to be on a super long bowsprit!

    Unfortunately mounting the wind sensor at the top of the mast makes shifting it from boat to boat very difficult, and junk masts wobble about a lot which could also affect the results.

    So making a 10 degree correction for downwash is probably the best option at the moment, especially for the data we already have.

    Other crazy ideas could include mounting the sensor on a drone!

<< First  < Prev   1   2   3   4   5   ...   Next >  Last >> 
       " ...there is nothing - absolutely nothing - half so much worth doing as simply messing about in junk-rigged boats" 
                                                               - the Chinese Water Rat

                                                              Site contents © the Junk Rig Association and/or individual authors

Powered by Wild Apricot Membership Software