Dynamic Pump Design of Complex Rising Main Injector Systems

Transcription

1 Dynamic Pump Design of Complex Rising Main Injector Systems Morten Just Kjølby 1, Arne Møller 1 1 DHI Denmark. (Presenter: Morten Just Kjølby mjk@dhigroup.com) Abstract In low laying coastal areas sewerage is usually conveyed in a system comprised of sewer pipe network, pumping stations and rising main networks before it enters the sewerage treatment plant. The rising main networks are often very complex with multiple pump injection points along each rising main line. The complexity of these systems provides engineering challenges in optimizing the design and operation of new and existing pressurised networks. A dynamic modelling approach is required to provide an overview of the hydraulics and water quality in new and existing pressurised networks. A specialised MOUSE PCS feature has been developed to facilitate dynamic pump design of pumps in new residential developments as well as facilitate analysis of pumps performances in existing pressurised networks. The MOUSE PCS (Pressurized Collection System) feature was developed by DHI in collaboration with Grundfos. The MOUSE PCS feature is available in the MIKE URBAN CS software package. This paper provides an introduction to the MOUSE PCS feature and the methodology applied in two projects undertaken by DHI and Grundfos in Australia and in Belgium. Keywords Dynamic hydraulic design, rising mains, complex injector systems, pump design and operation Introduction Optimal design or rehabilitation of pressurised networks and pumps is a common infrastructure and operational challenge for water utilities around the world. The optimal design of a pressurised network involves a broad range of concerns to be considered. The cost is likely to be the primary emphasis and includes cost of construction, operation, and maintenance for most water utilities. The initial capital investment is for infrastructure including pipes, pumps and operational equipment such as SCADA system. Operation and maintenance costs occur over time and include energy consumption to running the pumps and to the SCADA system. Due to climate change the emphasis on reducing the energy consumption has received increased focus lately in order to minimize the carbon footprint in the operation. The main constraints are that the sewerage must be conveyed through the pressurised pipes within a certain velocity range, overflows like SSO/CSO are not acceptable from the network as well as mechanical constraints to the operation of the pumps. Other requirements are related to water quality in terms of risk for formation of H 2 S which can lead to corrosion of infrastructure and be hazardous to the public health. In summary, the optimal design problem can be stated as: Objective: Minimize capital investment in infrastructure (pipes and pumps) and operational cost including energy consumption; 1

2 Subject to: Meeting hydraulic constraints to the pressurised and gravity pipe networks and pump operation, fulfilling variations in the load and meeting constraints related to the water quality in terms of risk for formation of H 2 S which can lead to corrosion of infrastructure and produce a public health hazard. A common design practice for pressurised networks is trial and error. An engineer selects alternative network and pump designs and usually these designs are based on common practice and simple hydraulic calculations. However with the increased complexity of the pressurised network with interactions between multiple pumps injecting into one pressurised pipe, identifying changes to improve a design can be a difficult task even for small networks. Especially if all the constraints listed above need to be verified and documented. Additionally to this it has become common practice by larger water utilities to put these design and rehabilitations works out in tender. As part of the tender hydraulic and water quality documentation of the design or rehabilitation is required documentation for suggested solution. A dynamic modelling approach is required to provide the designers and engineers with a complete overview of the pressure distribution, pump performance and water quality in the network in order to optimise the solution. Applied Methodology Objective The MOUSE PCS was developed by DHI in collaboration with Grundfos as a feature in the MIKE URBAN software package. The objective of the feature is to support the selection of pumps to meet project dependent design criteria. The tool provides a system wide overview of the hydraulics and the water quality in the pressurised network together with detailed information about the individual pump s performance. The selected pump configuration is evaluated statistically as well as system wide in terms of number of pumps running concurrently. The MIKE URBAN platform was chosen in order to achieve an industry acceptable modelling tool to meet the requirements in tenders around the world. Hydraulics In the design or rehabilitation process pumps are selected according to minimising the deviation between the specified pump duty point and the actual duty point as well as ensuring maximum efficiency. The duty point represents the intersection of the pump curve and the system resistance curve. At this point is not only the highest efficiency reached but also the point where velocity and therefore pressure is equal around the impeller. Operating the pump outside the recommended range causes an uneven pressure on the impeller which results in radial trust which deflects the shaft causing excess load on bearings, excess deflection of mechanical seal and uneven wear of shaft. A typical pump curve and system curve as well as duty point and recommended operating range is illustrated in Figure 1. Figure 1: Typical Pump Curve, System Curve and Duty Point 2

3 The sewer network is defined in MIKE URBAN and the MOUSE engine is used for dynamic simulations of the water levels, pressure, velocities and pipe and pump flows. As part of the simulation MOUSE PCS post process the pump results, pump curves and statistics is written automatically to EXCEL ensuring optimal usability in the design process or analysis of existing pump performance as well as to provide documentation. The individual pump performance is evaluated in a graph where the pump curve is plotted along with a histogram displaying how the pump has been running during the simulation. In a complex dynamic rising main network the pump is not running constantly at the duty point because of variations in the waste water loads during a day as well as the individual pumps are influenced by changes in the total head caused by other pumps in the system. When reviewing the pump results from the MOUSE PCS simulation the objective is therefore to ensure that pumps are running mainly within their operating range. If the pumps are running outside their operating range the selected pump is changed by another pump matching the operating head and flow better. The MOUSE HD Summary also reports operation times for each pump including information about if the pump has been forced by MOUSE to operate outside the pump curve. Information is provided for time of operation outside either end of the pump curve. The pipe velocities and pressures in the network are evaluated from the standard MOUSE HD simulation results to design criteria. Water Quality In pressurised sewer networks the risk of formation of H 2 S is a serious issue in terms of corrosion of infrastructure, odour issues and imposing a public health hazard. H 2 S is generated in the sewerage network under anaerobic conditions which occur in long rising main pipes or due to long retention times in un-ventilated parts of the sewer network. The problem with corrosion due to H 2 S usually occurs at the end of a rising main line when sewerage starts running by gravity but can also happen behind the pump due to too long standing water. It is important that the risk of H 2 S formation is evaluated at both the design and rehabilitation studies. The operation of the pumps plays an important role in minimizing the risk of H 2 S formation. The generation of H 2 S is the result of complex processes in the sewer network. Different mathematical models have been developed over the years. The MOUSE engine supports the empirical equations and the Z-risk factor approach as well as the extended WATS model. Both the Z-risk factor and the WATS model approach require a lot of input data such as characteristics of the waste water. In a design situation all this information may not be available. MOUSE also supports the water age concept which is traditionally used within water distribution modelling. The water age calculation can provide an indication of risks of H 2 S formation in the network based on water age. The development of water age is described by the advection-dispersion equation, where the first order decay term is substituted by a zero order growth term. Thus the one-dimensional vertically integrated equation for modelling the water age is given as where S = the age of the water; A = the cross section; T = the transport of the water that uses S; Q = the flow of the water; S S = the age of any source/sink; q = the source/sink; x = the coefficient; t = the time. (A S) + (T) t x = A + S s q 3

4 The applied methodology when using the MOUSE PCS in design or rehabilitation works is listed below: 1. Define desired standards of services; 2. Establish MIKE URBAN CS MOUSE model describing the network i.e. import from AutoCAD; 3. Define the pressurised pipe network and settings for MOUSE PCS; 4. Define the pipe material, head loss equation and roughness parameters; 5. Define pump stations, location, invert levels, start and stop levels and wet well geometry, 6. Select pumps, pump curves; 7. Define or assume the waste water load to the system and flow patterns. This could be ADWF (average dry weather flow) and PWWF (peak wet weather flow) = 5 * ADWF or other design alternatives; 8. Run Simulation 9. Evaluate results according to desired standards of services The Desired Standards of Services originates usually from the individual water utilities, provided in tender or given by industry standards within the country. Case Studies DHI has undertaken two projects in collaboration with Grundfos designing and analysing complex injector system as part of new developments in Cape Jaffa, South Australia and Kasterlee, Belgium. For each project a hydraulic network model was established to represent 21 pumping stations injecting into one rising main (Cape Jaffa, South Australia) and to represent the design of 102 pump stations injecting into a rising main system (Kasterlee, Belgium). The Kasterlee project was the most comprehensive and complicated of the two new developments to model and will be presented in this paper. Kasterlee Network Model In this project a MIKE URBAN network model was established to represent Kasterlee. This was done by importing the proposed network layout from AutoCAD and the network connectivity was established automatically using the post processing topology tool in MIKE URBAN. Invert levels and ground levels for manholes in the gravity sections and invert levels of fittings in rising mains were imported from text annotations in AutoCAD. The ground level of the fittings were set to top of rising main pipe in order to avoid introducing extra volume in the model. The network was made up by industry standard pipes with of sizes, DN63, DN75 and DN90. The diameters were converted to internal diameters in the model. HDPE 100 pipes was selected for the rising main sections and roughness was set to n = in the analysis. Diameters for manholes were assumed to be industry standard sizes and diameters of fittings were set to diameter of rising main pipes. The cover type of the manholes was set to Normal and Sealed for the fittings. The head loss at the manholes was set to 0.5 and the head loss at the fittings was set to no head loss. 102 pump stations were inserted with a size of 1700mm height and 1000*1000 mm. One pump was installed at each pump station. Invert level was set to Stop Level 0.2 m. Four different pump curves, SEG.40.09, SEG.40.12, SEG and SEG were supplied by Grundfos as suitable pumps in the Kasterlee network. A booster station was located in the middle of the network. The Kasterlee MIKE URBAN model is displayed in Figure 2. 4

5 Figure 2: MIKE URBAN Model of the area in Kasterlee Waste water loads were added to the model by assuming: Three properties connected to the each waste water manhole; One property (=1 ET) corresponds to 3 EP; One EP uses 150 l/ep/day. This means that the loading to each waste water manhole is 1350 l/day. The waste water load was assigned directly to each pump station. The number of waste water manholes per pump station ranges from 1-4 corresponding to 3-12 properties. Dry weather flow curves reflect the behaviour of different users populating an area. If the user category for a load is known then a user pattern can be multiplied on the load. Within waster water modelling typical users in an area can be categorised as Residential and Commercial. These profiles are recommended to be normalised so that their average value is 1. In this design job it was decided to apply a uniform profile to all loads. 5

6 Design Criteria The adopted desired standards of services relevant to this project are summarised in the Table 1. Table 1: Desired Standards of Services Item Design Criteria Value Sewage Flows: 1 Average Dry Weather Flow 150 l/ep/day (ADWF) Gravity Sewer Design: 2 Flow Equation Gravity Sewers Manning s 3 Manning s friction factor n Plastic, n = Minimum Velocity C1 factored Flow 0.6 m/s Sewage Pumping Station Design: 5 Wet Well Storage Requirements (m 3 between pump start and stop levels (0.9 * Single Pump Capacity) / N Where: Single pump capacity measured in l/s, N = 10 starts per hour for < 50 kw motor and; N = 6 starts per hour for > 50 kw motor. 6 Single Pump Capacity C 1 * ADWF C 1 = 15*(EP) (C 1 Range 3.5 to 5.0) 7 Total SPS Capacity 5* ADWF or Design Flow 8 Emergency Storage 4 hours at ADWF (local gravity catchment) Rising Main Design: 9 Flow Equation Manning 10 Manning, n HDPE, n = (ks = 0.15mm) 11 Minimum Velocity (single pump) 0.7 m/s 12 Design Velocity 1.0 to 1.5 m/s 13 Maximum Velocity (single pump) 2.0 m/s 14 Maximum Velocity (all pumps) 2.5 m/s n = (ks 1000)1/ Simulation Results and Discussion Hydraulic analysis of the proposed pump capacities and operation was undertaken during different design flow scenarios. The pump operation during average dry weather conditions (ADWF) were analysed adopting design criteria that pumps were not exceeding a maximum number of 10 starts per hour and no pump dry stops. The MOUSE PCS feature post processes the MOUSE HD simulation results to produce pump operation histograms as well as producing various statistics for each pump. A sheet for each pump can be produced as displayed in Figure 3 by selecting the pump. 6

7 Figure 3: Pump Results At the left hand side of Figure 3 pump station configuration information can be found as well as simulation statistics such as number of pump starts per day, averaged volume pumps, minimum and maximum velocities and flows in the rising main. At the right hand side of Figure 3 the pump operation histogram together width the pump curve is displayed. The pump curve for each pump is chosen on the basis of the information about the pump operating on the curve as well as the histogram displayed in. The aim is to get the pump to operate around the duty point most of the time. It is a dynamic simulation that is undertaken with different pumps in operation at different times during the simulation. It is therefore not possible to provide a single duty Q and H value per pump. After some iteration in the selection of pumps almost all 102 pumps are operating on their curve during the whole simulation period. Pump Summary Discharge and Operation The pump summary table from the MOUSE HD simulation provides information about minimum, maximum pump discharge, number of pump starts, number of pump dry stops, accumulated flow, speed, operation total and operation outside pump curve. Pump dry stops happens when the available volume + the inflow volume to the wet well is less than the pumped discharge during the pump shut down period. If the wet well runs dry then the pumps are shut down immediately by the model. It is not acceptable to have any pumps in the model forced to be closed down. With only 10 centimetres between pump stop level and invert level of wet well a few pumps will be forced to be closed down by the model. For this reason it has been requested that the pump stop level is 20 centimetres above invert level for all wet wells. 7

8 During the simulation it might be required that MOUSE operates outside the pump curve which can be set up to be done automatically. In the pump summary the operation time and operation time either above or below the pump curve are reported. This information is utilised together with the MOUSE PCS results in written to the EXCEL to select the right pump curve for the wet well. The aim is to reduce the operation time outside the pump curve as much as possible so that the pump is operating at the curve as well as not breaching the desired standards of services. A few pumps were operating for a short time outside their pump curve. The operation is below their curves indicating that some of these pumps may be reduced in size. The number of pump starts per day was not breaching the desired standards of services. Rising Main Velocites The maximum velocities range from m/s to m/s. Only 3 pipes have velocities above 2 m/s. The maximum velocities in the rising main pipes are not breaching the desired standards of services of maximum 2.5 l/s. Pumps running concurrent Statistics of pumps running concurrent is produced in the EXCEL spreadsheet. As presented below up to 6 pumps are running at the same time upstream the booster station. During 96.7 % of the time maximum three pumps will be running concurrently. Number of pumps Percentage of time Accumulated Percentage running concurrent [%] of time [%] Table 2: Pumps running concurrent During this project it was discussed that only two pumps can be in operation at the same time. In the model all the pumps are controlled locally by start and stop levels. It would require system wide control of the pumps to ensure that only two pumps are running concurrently. This is currently not supported in by MOUSE PCS feature. The reason for controlling the number of pumps is to reduce the operating head range required for the pumps and thereby reducing the investment and operating costs. The table showing number of concurrent running pumps also indicates that in 54.1 % of the time no pumps are running. This also indicates that it will be possible to operate the system with a maximum of two pumps running at the same time. This operation may be achieved by substituting the fixed pump start level by a range for the start level combined with a sensor for measuring the actual pressure head. If the head is too high then the start of the pump should be delayed until other pumps are stopped and the head drops. This type of operation could be tested with the dynamic simulations with a few additional features added to the computations. Water Quality The purpose of the water age simulation is to provide an indication of risk of H 2 S formation. The water age simulation is not a H 2 S simulation but can be used as an indicator under the assumption that the older the sewage gets in the pipes the higher risk of H 2 S formation is expected. In Figure 4 is shown the water age results in the network upstream of booster station P86. 8

9 Figure 4: Water Age simulation results north of P86 The maximum water age in the network is 3 hours. The results indicate that the risk of H 2 S formation is highest upstream P13 and P27. Future Development During the project ideas for future development has been identified: Enhanced support of model control of number of pumps running concurrent; Intermediate sections of gravity pipes needs to be supported by MOUSE PCS. This is supported in the standard MOUSE HD; Improved support for cost options; Improved support for minimising operation costs due to energy consumptions and thereby minimising the Carbon Footprint (if the energy is produced by fossils); Conclusion The MOUSE PCS concept has proven to be a very useful concept in these two projects when applied in design works of complex rising main networks. The introduction of dynamic simulations in the design phase provides the designer with a much better overview of the velocities and pressures in the network at any time. The complexity of pumps running concurrent and injecting into one rising main is impossible to evaluate unless running dynamic simulations. The water utilities requests proof of design from the designers and manufacturers that the network and pump configuration operates as expected. The option of running water quality simulation to determine risk of H 2 S formation is also very valuable for the client or water utility. 9