« profile & posts archive

This author has written 591 posts for Larvatus Prodeo.

Return to: Homepage | Blog Index

16 responses to “Could the Wivenhoe flood engineers have done better?”

  1. fmark

    FYI, mods: The “other post” link returns me to this same page.

  2. Peter Smith

    Thanks for another great analysis, Brian.

    I’m glad to be retired now. I don’t recall so much hassle from lawyers when I was in the field – modeling the 1974 flood (in retrospect).

    Just one thing; Towwomba? Must have been a late night. 🙂

  3. Roger Jones


    after your first post I read some of the report. They spent ages speculating about what was in the engineer’s minds and less on how the flood was managed. The amateur psychology is speculative bullshit.

    Because it is a state government report, they can’t make recommendations for the Commonwealth but it makes sense to have a live storm-flood modelling capacity run by the Bureau. The judgement calls by the engineers were all made because they had little solid info about how much rain was coming. It would pay for itself easy.

  4. Chris

    Roger @ 5 – is it possible at the moment to do modeling accurately enough and fast enough to be useful?

  5. marks

    Given that there is no reason why the coming summer should not see a similar flood, the points you raise are important to be resolved well before then.

    If such a flood occurred, how would dam operators operate? Would they strictly observe the manual even if a better strategy presented itself? More important, without resolution of the questions you have put up in these two posts, who would be silly enough to take the post of operating the dam? I say silly, because if operating the dam according to the manual is not optimal under some conditions, and if use of judgement is likely to end one’s career, who would take it on? At the moment, this seems to be the case.

    If the new government in Queensland is not careful, the end point of all this might be that the next flood might end up being managed by people less qualified or able than were managing the last. Not at all a pleasing prospect for those living downstream of the dam.

  6. Roger Jones


    it seems that even if they had a better idea about the severity of the rainfall they would have increased outflows earlier (there’s a certain amount of hindsight colouring the report).

    A high-resolution weather model will pick up storms better, good computing power allows more ensembles to be run. Coupled to a regional flood model it can give an ok idea of where particular storms may hit hardest. There wouldn’t be accurate enough predictions to take away the doubt, but the capacity is useful for multiple events, especially when you’ve got more rain on a flood, and for managing peaks.

    It’s expensive and data hungry.

  7. John D

    Looking at the first graph it is clear that, with hindsight, the ultimate peak could have been reduced if flows had been increased sooner than what was actually done. Some of this increase could have consisted raising flows to just below the point where key bridges were blocked or raising flows to levels to just below the start of urban damage.
    If we look at the graph from the viewpoint of what the engineers would have known at any time it could also be argued that there is no real reason why below bridge block flows should have started as soon as dam level was much above the end of wet season level target. (Do you know what this is in terms of elevation Brian?) It could also be argued that moving to block the bridge flows sooner would have been worthwhile given that this means trading some inconvenience against the potential reduction in damage if the situation deteriorates. Keep in mind here that, because of the shape of the valley, the amount of water stored between an elevation of 67 and 68 is much much less than that stored between 74 and 75, ie, holding the dam level lower during the early stages would not have made much difference to ultimate dam level reached or peak flow.
    Increasing flows to above the urban damage point sooner would have made some difference but it worth noting that the flows actually used would have been adequate if the second major rain event hadn’t occurred. Given how close the dam came to blowing the first fusion plug it would have been very risky to have started restricting the flow from the dam before the peak outflow was reached.

    One of the points highlighted in the post was the uncertainty of when urban inundation damage would actually start and what the relationship between level and damage is. It may be that lifting a small number of houses and/or putting in a small number of low levee banks would allow the dam to discharge at higher levels without having to worry about the damage this would cause.

    If you haven’t already done so it is worth reading what Desipis @31 had to say re the manual setting priorities at various stages while leaving it up to the engineers to work out what has to be done. It is also important to understand that at any stage, the engineers should be trying to achieve lower stage priorities as well as the top priority for the stage.

  8. Roger Jones


    A few years ago I saw simulations of a system that Alan Seed of the BoM was trying around Melbourne. A high-res storm system on a fine model grid has a chance of picking up some of these. I’m thinking of the benefits at the regional scale – don’t know how the local storm cell benefits at the Lockyer Valley scale would play out.

    But even before the event, high-res runs would be handy. Back-casting will be done eventually, but some of these falls would have been getting into probable maximum precipitation territory.

  9. John D

    Brian: The table in chapter 9 provides a lot of detail to play with. I won’t bore you with all the calcs I made but a few things do stand out:
    I have used the first figures in the table (Figures @ 0900hrs, 6/1) as the base. At this point the dam level was 0.32 metres above “full”
    1. The gates started opening @ 1500 hrs, 7/1 – 28 hrs after base with the dam holding 79 GL more than base with dam level @ 68.03 m.
    2. Total outflow reached the 2000 m3/sec bridge limit @ 0900 hrs 10/1 – 96 hrs after base with volume 520 GL above base, level @ 71.58 (The first fuse plug opens at 75.7 m)
    3.Total outflow reached the 3000 m3/sec @ 0900 hrs, 11/1 – 120 hrs after base with volume 847 GL above base, level @ 73.81 m.
    4. Peak inflow (11, 561 m3/sec) was reached @ 1300hrs, 11/1 – 124 hrs after base.
    5. Peak level (74.97 m) and peak out flow (7464 m3/sec) were both reached @1900 hrs, 11/1 – 130 hrs after base. Not sure whether the gates were fully open at peak flow.
    6. It will take about 155 GL to raise dam level one meter starting at 73.5 m.
    7. 1000 m3/sec=3.6 GL/hr.
    8. The difference between peak inflow and peak outflow was 4097 m3/h. At this net inflow the dam would have would have taken 7.7 hrs from the peak level to the point where the first fuse plug would have opened.

    All the above suggests that gains would have been made if the outflow was set just below the bridge close level while the dam was above the nominal full level of 67 m. For example, the peak level would have been about 3.4 m lower if all outflows above 2000 m3/sec had remained the same. Given how close the dam was to reaching the point where the first fusible plug would have operated it is debatable whether the engineers would have used the gains to reduce the peak flows.

    It is worth noting that reducing dam level to 75% at the start of the wet season only gives 300 GL of extra space.

  10. John D

    Brian: In 20/20 hindsight it is obvious that peak flood levels would have been lower if higher flows had been used earlier. For example, a visual on your first graph suggests that, if outflows had been set a bit below 4000 m3/s as soon as dam level reached 70 m, this would have been sufficient to avoid reaching fusible plug level and there would have been no need to go to the peak out flow of 7464 m3/sec. This would have reduced significantly the number of properties damage and the extent of damage to other properties. Some lawyers would see this as a basis for making claims against the dam operators on the grounds that their clients properties were damaged by the dam operators decisions.
    The problem with the above analysis is that having a policy of going to 4000 m3/s would have caused necessary damage if the weather had matched the forecast and the second big surge of rainfall had not occurred. In this case less that 3000 m3/s would have been required and the crisis may have been avoided without the need for flows that would cause urban damage. If the operators had gone to a peak of 4000 m3/sec under these circumstances there would have been another group of property owners claiming damages because of dam operator decisions.

    As I have said before, the manual should be changed to encourage dam operators to lift flows to levels that will not cause damage much earlier in a potential flood event. However, it needs to be clearly understood that there will be very rare occasions when flows will have to lifted to higher levels and these decisions will need to be made by expert operators, not manual operators nor politicians.
    There is also a need to clarify just what levels actually do cause damage and the extent of damage at higher levels. Also a need to consider spending money to increase the volume of “no damage” flows.

    In the meantime, LNP blocks SEQWATER statement refuting some of the commissions findings Government blocking could render SEQWATER insurance invalid.