Bentley Hammer - Combination Pump Curve

I cannot adjust the pumping system configuration as needed. My system consists of pumps arranged in series and parallel.

The total flow of the pumping system is 1400m³/h, with a pressure of 37mwc, distributed as follows:

  1. In series: PMP-1a (467m³/h, 21 mwc) and PMP-1b (467m³/h, 16 mwc), resulting in a total pressure of 37mwc.
  2. In series: PMP-2a (467m³/h @ 21 mwc) and PMP-2b (467m³/h, 16 mwc), also resulting in 37mwc.
  3. In series: PMP-3a (467m³/h @ 21 mwc) and PMP-3b (467m³/h, 16 mwc), with the same result of 37mwc.

The parallel system of pumps 1, 2, and 3 also produces a total flow of 1400m³/h, with a pressure of 37mwc.

The flow and pressure values I mentioned for each pump indicate its contribution to the system. However, the result calculated by the program does not match the expected sum of flows and pressures. How can I perform this combination in Hammer, since this curve merging option is only available in WaterCAD?

  • Hi Danilo,

    Pumps have a Duty Curve consisting of an infinite number of Duty Flow vs Head points.   WaterCAD has estimated all other Duty Flow/Head points represented by the Blue Curve based on the single Duty/Flow point value entered.

    I would tend to recommend though not using 1 Point/Estimated Curves in Detail Design/Analysis however.  Use the Curves published by KSB in their Trade Catalogue.  Contact their local sales rep to get the Catalogue (who locally supplied ours to us for our fleet of KSB pumps), and input these as Multiple Point curves.

    So, because of the infinite possible pumped flow/heads in pumped systems,  the actual model/real-life duty flow in the pumps is variable.  At any given time, the actual delivered flow is at the point of intersection between the effective pump curve and the system curve.  The effective pump curve will vary with the number of parallel and series pumps on or off, and the system curve varies based on upstream available head, downstream demands, discharge tank HGL etc.

    ie. The model will only output a modelled pump flow rate equal to an input Duty Flow/Head if by pure chance the system curve intersects one of the input Duty Flow/Heads points.  By this will only be by coincidence, in virtually all real-life design scenario the system curve will not intersect with a value the modeller has entered, since the true curve has infinite possible duty points and the pump operates where the system curve intersects.

  • Hi Ben.
    Thank you for your feedback, I agree with the use of a curve with more points, but the question is how should I treat this case that I mentioned above, how should I construct each curve so that the system understands it.

  • In Hammer, you generally will want to keep the pumps modelled separately where possible, and not create any merged pumped curves.   The key reason for that is generally a Hammer transient model scenario on pumped systems is for pump station power failure. To accurately calculate this Hammer will likely need the Moment of Inertia of the combined mass of the rotating pump impeller and motor (this information needs to be entered in the "Transient" Tab for the Pump Curve for these Pump Off/Power Failure scenarios).  This becomes more difficult to simulate properly with a "combined" Moments of Inertia are not easily additive, hence modelling as individual pumps is the default method.

    Modelling pumps individually is also the default WaterCAD method as well, in most cases combining pump curves is unnecessary. For our utility base models we have not had to do that in any of our ~150 modelled pump stations.

    However, even if a merged curve representing combined parallel/series pumps was input, this won't change the effective model pump curve, and won't change the system curve.  The model will still calculate the same overall duty flow.  The result you are getting in Hammer Steady State is the same result you will get in WaterCAD and the same result if the pumps are modelled as a single virtual pump.

    Looking at some attributes that are visible in the model some checks may be in order though to ensure that the correct System Curve is being input. There are indications that some inputs may be causing the model to have System Curve with  pipeline head losses that are too small, and is potentially why the 3 x parallel pumps are operating at a modeled duty flow significantly higher than expected:

    • Noted is that the Rising Main Material has been input as "Steel".  If the pipeline is Mild Steel, then note their manufacturing process results in some differences in how their Nominal Diameter (DN) value is used.  Mild Steel is an unusual pipe material in the way it gets made.  It is cold rolled into a welded tube from plate steel fed into the roller at an adjustable, custom angle that is set according to what outside diameter tube the customer ordering the tubing wants. The customer buying the pipe specifies the Outside Diameter by selecting from a manufacturer's "DN" Diameter technical data sheet.  Mild Steel is then one of the few materials where the typical "DN" used on design drawings or contract specifications is nowhere near the actual internal diameter of the pipe.  Be careful here that the model "Diameter" values input are the actual internal diameters of the Mild Steel pipeline.  A 520 Outside Diameter Mild Steel pipe will instead be closer to an internal diameter of 450 mm with a cement liner, and this is more the size of pipe I would expect for matching standard DN450 or 18'' Ductile Iron fittings and valves bolted or welded at the joints.  These can generally only be found by looking up the internal diameters for a given "DN" in the manufacturer's Trade Catalogue
    • Noted is the extremely high velocity of ~3.0 m/s.  In this size pipe this indicates the modeled length of the rising main has been input as a relatively short length.  The first thing to check is that the length of pipeline has been input correctly.
    • If the rising main is short, then care is needed to accurately model the head losses in the pump suction and discharge manifolds.  The minor loss coefficients for all bends, tees and valves need to also be input into the Pipe Minor Loss Coefficient attributes where these can become quite significant in the calculated System Curve since in short, large diameter rising mains there is otherwise very little pipe wall area to cause any flow resistance.

  • Ben,

    I will check my model, the design velocity is 1.92m/s, for a flow of 1400m3/h, is the diameter that I have to enter in this field the internal diameter?

    before                           after


    Regarding the steel tube ASTM A 36 galvanized tube, I used the Steel data available in Hammer.


    Thank you

  • Yes, this is Internal Diameter.

    For different reasons, PE mains are also often listed in their technical datasheets with a "DN" equal to their Outside Diameter  .  From the above table, entering a value of "315" for DN315 PE will not be the right value. This is due to the most common PE manufacturing standards being based on ISO Standard ISO 4427, in which ISO convention was to use the Outside Diameter as the dimension for  "DN".

    The wall thicknesses need to be subtracted in order to determine the correct  model Diameter.

    Inside the transient modelling runs, additionally the Young's Modulus (Elasticity) and Poisson's Ratio (Bulk Modulus) of the Pipe Material be needed along with the effective pipe wall thickness in order to calculate the transient pressure wave(s) celerity.