Intelligent Sizing of Detention Basins Using a Dynamic Hydraulic Model

Transcription

1 Intelligent Sizing of Detention Basins Using a Dynamic Hydraulic Model M Sharkey 1 1 Bentley Systems Pty Ltd, mal.sharkey@bentley.com Abstract Stormwater detention basins are an important part of Water Sensitive Urban Design (WSUD). When designing such a structure it is critical to size the basin correctly, but the problem of determining the correct size for a detention basin is far from trivial. Fortunately, purpose-built hydraulic modelling computer programs can perform much of the computational heavy lifting allowing engineers to make a better estimate of required detention basin size. This paper investigates the use of one such hydraulic model (CivilStorm by Bentley) in the design of a detention basin (and associated appurtenances) in a hypothetical urban design scenario. This paper demonstrates how the model converts rainfall into catchment runoff, routes the runoff through a stormwater system and then presents the results to the engineer. This paper then shows how it is possible to optimize the size of the detention basin for the particular system and design event using output from the model. Finally the paper will highlight some of the possible consequences of sizing a basin incorrectly. Introduction Stormwater detention is the temporary storage of stormwater runoff in ponds, depressions, basins or underground containers. (For the purposes of this paper we will refer to these structures collectively as stormwater detention basins though the discussion is equally applicable to any stormwater detention structure). These detention basins are an important part of Water Sensitive Urban Design (WSUD) because they typically detain the runoff for some time to reduce the peak runoff rate, while allowing some (or all) of the runoff to infiltrate slowly into the ground to reduce the total runoff volume. This is a useful stormwater management technique at the site of new urban developments, because the increased imperviousness of the developed area (after construction of streets, driveways, roofs, etc.) increases the runoff rate, which, if left uncontrolled, typically increases the severity and frequency of flooding downstream. By controlling this excess runoff with a detention basin the effects of the increased runoff can be mitigated, the natural hydrologic behaviour of the surrounding waterways can be maintained, and the basin can also then be utilized to improve runoff quality by allowing sediments to settle out, etc. However one of the main difficulties with utilizing a detention basin is in the design, because calculating the appropriate size for the basin (and connected appurtenances) so that they can handle the flows directed to them is a very demanding computational task. Yet it is critical to size these basins correctly: if they are too small there is a chance that they could overtop in a severe storm potentially damaging the basin structure or causing flooding in nearby urban areas; but if they are too big, they may be unsightly, expensive to construct, use too much potentially developable lane, and may also retard natural environmental flows. When designing the basin, factors such as the basin capacity and configuration, the capacity of the stormwater conveyance system upstream and downstream of the basin, the capacity of the basin outlet structure and the predicted design runoff hydrograph should all be taken into consideration. It quickly becomes obvious when attempting to perform these calculations by hand or in a spreadsheet that, at best, it will only be possible to arrive at a coarse approximation. With the aid of a personal computer, and a purpose-built hydraulic modelling

2 software application the problem becomes much more manageable, and allows for better, more economical and efficient designs. This paper investigates the use of one such hydraulic model (CivilStorm by Bentley) in the design of a detention basin (and associated appurtenances) in a hypothetical urban design scenario. The paper begins with a brief overview of stormwater detention. Next the paper demonstrates how the model converts rainfall into catchment runoff, routes the runoff through a stormwater system of inlet pits, pipes, open channels, detention basins, detention basin outlet structures and culverts, and then presents the results to the engineer. The paper then shows how it is possible to optimize the size of the detention basin for the particular system and design event using output from the model, and finally the paper discusses some of the limitations of existing methods and tools for sizing detention basins, and highlights some of the possible consequences of sizing a basin incorrectly. Overview of Stormwater Detention Figure 1 illustrates the basic concepts involved in stormwater detention analysis. An inflow hydrograph from one or more contributing drainage areas is directed to the detention basin. Runoff is then released from the detention basin at a controlled rate via the detention basin outlet structure (typically comprised of one or more weirs, culverts, risers or orifices in any combination). Figure 1 - Conceptual Drawing of a Detention Basin (Haestad Methods et al., 2003) The controlling effect of the outlet structure acts to attenuate the peak outflow discharge rate, as well as increase the time to peak basin outflow (relative to peak inflow). The area between the basin inflow hydrograph and the basin outflow hydrograph (during the time when inflow is greater than outflow) defines the volume of detention storage required for the system (see Figure 2)

3 Figure 2 - Basin Inflow and Outflow Hydrographs (Haestad Methods et al., 2003) Rainfall Runoff and Hydraulic Routing Calculations The basin inflow hydrograph is a function of the runoff hydrograph(s) from the upstream drainage area(s), as well any attenuation of the runoff as it is routed through the stormwater system upstream of the pond. A detailed analysis of the various methods of computing stormwater runoff is beyond the scope of this paper, however a good hydraulic modelling application should provide multiple methods for computing the runoff after a user enters the required data, such as rainfall information, runoff area and infiltration loss information. The model should also allow users to enter a user-defined hydrograph which was computed elsewhere (e.g. a spreadsheet or another application). The routing calculations in the conduits of a stormwater system can be handled using either hydraulic or hydrologic routing. Hydrologic methods (such as the Muskingum method) involve the principle of conservation of mass and a simple storage-discharge relationship (Chow et al., 1988), while hydraulic methods involve the simultaneous solution of the Saint- Venant equations (Haestad Methods et al., 2003). The hydraulic methods analyse both flow and hydraulic grade line, and are particularly well suited to complex systems with a mix a free-surface and pressurized flow, as well as flooding and ponding at structures (Jin et al., 2002). The CivilStorm model (Bentley, 2007) is based on the full hydrodynamic solution of the onedimensional unsteady flow (St Venant) equations. This model uses a four-point implicit finite difference scheme that has the advantage of having good stability for large computational time steps while maintaining good accuracy. It also exhibits robustness in modelling very complex situations, such as interchanges between open channel gravity flows, pressure flows, and storage effects from street flooding, bifurcated pipe networks, flow regime changes between subcritical and supercritical conditions, etc. (Jin et al., 2002). As such, it is a good choice for modelling a range of different stormwater systems. The basin outflow hydrograph is dependent on the basin outlet structure, as well as the capacity of the conveyance system downstream. The hydraulic model should be able to compute the flow-elevation rating curve for a complex pond outlet structure, and use that rating curve in the flow routing calculations. CivilStorm allows users to define complex composite outlet structures (made up of a combination of weirs, orifices, risers and culverts), which provides the flexibility to handle a

4 wide range of real-world situations (Bentley, 2007). For example, Figure 3 shows a composite outlet structure containing an inlet box, a culvert, a V-notch weir and a spillway Figure 3 - Composite Outlet Structure (Haestad Methods et al., 2003) Once these structures are defined by the user, CivilStorm will compute the composite rating curves, then route flow from the pond accordingly. Figure 4 shows a typical elevation-flow relationship for the structure in Figure 3. CivilStorm will compute basin discharge in accordance with this relationship. Figure 4 - Composite Outlet Structure Rating Curve (Haestad Methods et al., 2003) In addition to the elevation-flow curve show above, CivilStorm can compute basin outflow for variable tailwater levels (where the water downstream of the basin varies due to the downstream system capacity or tidal conditions). The tailwater level can affect the flow out of the basin, and can possibly even cause backflow into the basin. Basin Sizing Considerations and Optimization of Basin Size As discussed in the previous section, the required size for a detention basin depends on the basin inflow and outflow hydrographs. However there are generally other considerations as well such as: How much land is available to construct the basin?

5 Are there local regulations that stipulate a minimum detention volume amount (such as a first-flush requirement)? Is the site accessible for construction and ongoing maintenance? What are the local geotechnical conditions? Can the side walls be steep, or is there risk of collapse? Will there be significant infiltration into the soil? Are there local regulations that stipulate that the peak runoff rate or volume from a developed site must be within a certain tolerance of the peak runoff rate and volume prior to development? Of these, the most problematic as far as the hydraulic calculations are concerned is the final point in the list. Due to the interconnected nature of most stormwater system, constraints such as those where the peak runoff rate and/or runoff volume must be close to the predevelopment condition are difficult to achieve without significant trial and error. Fortunately this trial and error is simple in tools like CivilStorm which support scenario management (where several what-if trials can be set up in the same model file to make comparison of different trials quick and easy). As an example of this basin sizing process, we will consider the system shown in Figure 5. Figure 5 - Hypothetical Stormwater System in CivilStorm In this system we are considering the re-design of the stormwater system to support the development of a new office complex, parking area and access road. The system consists of the following components: 3 detention basins (2 existing, and one proposed as part of the new development), each with an outlet control structure. 500m of proposed underground stormwater mains 1500m of existing natural and man-made drainage channels 32 separate sub-catchments

6 12 proposed stormwater catch basins 3 existing culvert structures A summary of the calculation details used in this model are given in Table 1 below: Table 1 - CivilStorm Calculation Details Calculation Component Calculation Method Catchment Loss Method Horton Rainfall Event 100 Yr ARI, 24 hr Duration using time-depth data a 6 min increments Runoff Calculation Method Unit Hydrograph Conduit/Channel Flow Routing Hydrodynamic solution of the one-dimensional unsteady flow (St Venant) equations Detention Basin Flow Routing Numerical Integration using 4 th order Runge- Kutta approximation Culvert Capacity Analysis Based on U.S. Federal Highway Administration HDS No. 5, Hydraulic Design of Highway Culverts nomographs Basin Outlet Capacity Calculation Combination of weir and orifice equations An initial simulation which compared the site runoff prior to development to the site runoff after development gave the following results: Note that in this case the red line represents the runoff after site development, but the postdevelopment scenario results above do not include the effects of the new detention pond. As shown, the peak runoff is approximately 45% higher after development than before development, and the time to peak occurs approximately 30 minutes earlier. This has the potential to adversely affect the stormwater system downstream. Figure 6 - Pre- vs. Post Development Runoff Results with no Additional Flow Control

7 Next we introduce a new scenario into the model which contains a detention basin (2m deep, surface area 0.17 ha), in an effort to reduce the peak. The basin uses a simple riser for its outlet structure. Again the dotted line shows post-development runoff: In this case the pond and outlet structure successfully lowered the peak outflow, however it is now 45% less than the predevelopment peak. Figure 7 - Pre- vs. Post-Development Runoff after Including a Detention Basin This structure could be further optimized to possibly reduce construction costs (by making the basin smaller), and also to better match existing natural flow conditions. So, after further design iterations, the basin size is reduced to 2m deep with a 0.85ha top area. The basin outlet structure is augmented with a low flow orifice, as well as a high flow weir spillway. The results are show below: The peak flow now matches the predevelopment peak to within 3%. In addition the peak flow times are within 15 minutes of each other. This is adopted as the final design. Figure 8 - Pre- vs. Post-Development Runoff after Optimizing Basin and Outlet Structure

8 Limitations of Existing Methods and Tools for Sizing Detention Basins There are a range of existing methods and tools used for sizing detention basins, ranging from simple back of the envelope type approximations through to complex flow routing using numerical integration. Table 2 (below) summarizes some of those methods/tools, as well as their limitations: Table 2 - Limitations of existing methods and tools for sizing detention basins Method Description Limitations Rule of thumb Storage size determined based on Complex hydraulics will mean no two (guess) previous experience. detention basins will behave the same, so Estimation Methods Storage- Indication (a.k.a modified Puls or level pool routing) Method Numerical integration Using a known inflow hydrograph, estimate the outflow hydrograph and find the volume difference between them this is the estimated basin size. Using a known basin stage-storage curve and outlet stage-discharge curve, compute a storage indication curve. Then use the storage indication curve and inflow hydrograph to compute outflow at discrete time steps. Solve the conservation of mass equation in the form: dh = I (t) O (h) dt A(h) (Where: h = stage; t = time; I = inflow; O = outflow; & A = water surface area) using accurate numerical integration. this method is not reliable. Need to design pond and outlet so outflow matches estimate. This is not trivial, and if not done correctly, the basin size could be inadequate. Difficult to manage the calculations by hand need a detailed spreadsheet or computer model. This method doesn t consider the effects of the stormwater system upstream or downstream (for example, time-varying tailwater levels), which can influence the inflow and outflow hydrographs. Again the calculations are difficult to manage without a spreadsheet or computer model. If the model does not take into account the effects of the stormwater system upstream or downstream, it may not be accurate. CivilStorm overcomes these limitations by performing Numerical Integration using 4 th order Runge-Kutta approximation, and allowing for the effects of the upstream and downstream system at each timestep when determining basin inflows and outflows. This means that complex situations such as interconnected basins or networks with significant surcharging (where basin outflows can be reduced considerably, or even reversed) can be modelled accurately. Consequences of Sizing a Basin Incorrectly The process of sizing a detention basin is important because: If the basin is too small it may overtop in severe storms, possibly damaging the basin structure or causing flooding downstream If the basin is too large it may be unnecessarily expensive to construct, may be unsightly (think of a large water feature that is permanently dry), or may retard natural environmental flows in the drainage system If the basin is not analysed and sized in conjunction with the stormwater system it is connected to, it may actually cause an increase in the peak runoff rate. This last point is counter-intuitive and so warrants further explanation. Consider the system show in Figure 9. The system shows two tributaries, A and B. A detention basin is to be constructed for a development site on tributary B. The graph show the pre-developed

9 hydrograph, as well as two post-development hydrographs (one for system B with a detention basin, the other without). Figure 9 - Effect of Detention Lag on Overall Basin Hydrograph (Haestad Methods et al., 2003) As shown in the graph, the detention basin successfully attenuates the peak post-developed flow from tributary B, however flow is still being released from the basin as runoff from the larger drainage area peaks increasing the peak runoff for the watershed as a whole. In addition, flow lag due to detention may increase the time that erosion occurs in the downstream In the above example, the peak flow increase would be difficult to detect without utilizing a model of the entire stormwater system. Summary Detention basins form a part of many Water Sensitive Urban Designs. This paper points out that the hydraulic analysis and correct sizing of these structures as well as the stormwater system they are connected to is an important, but complex part of the design. However, designers should be aware that specialized software tools such as CivilStorm by Bentley can assist with the hydraulic analysis of their systems, which may translate into time savings during the design process and/or the creation of more suitable designs. References BENTLEY (2007) CivilStorm. Exton, PA, Bentley Systems Incorporated. CHOW, V. T., MAIDMENT, D. R. & MAYS, L. W. (1988) Applied Hydrology, New York, McGraw Hill. HAESTAD METHODS & DURRANS, R. (2003) Stormwater Conveyance Modeling and Design, Haestad Press, Waterbury CT, USA. JIN, M., CORAN, S. & COOK, J. (2002) New One Dimensional Implicit Numerical Dynamic Sewer and Storm Model. 9th International Conference on Urban Drainage. American Society of Civil Engineers (ASCE).