Dynamic Modelling of Control Valves Observed PRV d/s Pressure Against Flow Pressure (m) 28 27.5 27 26.5 26 25.5 25 24.5 24 0 0.5 1 1.5 2 2.5 Element Flow (l/s) Andy Rowland Technical Lead Water Networks
Background 2016 Conference Workshop started a detailed look at the modelling of PRV s. Thoughts that came to mind at the time: Focus is often on model calibration under normal flow conditions. BUT Models are frequently used to identify and resolve system capacity issues. INVESTIGATE MECHANISMS INFLUENCING PRV OUTLET PRESSURE Growth / Resilience / New Developments / Fire Flow / Incident Management Use and influence of K / Cv values consequences of improper use Why typically, is low priority given to this value that essentially quantifies valve capacity Fixed outlet PRVs that are performing sub-ideally must be responding to external influences and can t be altogether random! Could the valves also be responding to flow and / or upstream pressure influences in addition to the ideal control setting?
Dynamic Modelling Accurately modelling the response of assets to external influence Delivering: Increased model confidence across broad range of demand scenarios Which is important when: Understanding and improving network resilience / the ability for the network to absorb unplanned events. Operational response modelling Planning and designing for growth / high demand scenarios Network optimisation and energy management
Current Standard Practice For PRV s (but applicable to most control valves) * Level of widespread use Capacity Correct Diameter Use software default k Manufacturers wide-open k Use achievable max-opening k Modulation (responsiveness) Fixed outlet element for PRVs without an external modulation controller Flow / time modulated elements to model actively modulated PRVs. Fixed throttle to simulate d/s variation Variable modulation elements to handle uncontrolled valves with d/s variation
Representing the MODULATION function of a Control Valve To throttle or not to throttle: Head-loss induced across a throttled valve is a function of flow ARE varying outlet pressures Head-loss increases exponentially as a clear function of flow Introducing a throttle downstream ACTUALLY of the control valve we inadvertently model a flow-modulated downstream pressure. a function of Flow? Synergi VQ and InfoWorks FMV control valve elements both have the facility to define Flow Influenced outlet pressures:
Representing the MODULATION function of a Control Valve 80mm Cla-Val NGE - fixed outlet Observed PRV d/s Pressure Against Flow 28 VOLATILE low) FLOW Notice (but with throttle 27.5 Notice rate of pressure Rate of Pressure Reduction Similar Example Notice reduction reduces as flow flow increases with increased in InfoWorks Extrapolation Profiles Observed increases 27 Variable Pressure Downstream of a PRV most likely indicate INLET /HEAD 72M - STABLE Calibrated Dynamically Responsive PASSIVE FLOW MODULATION Pressure (m) Example: 26.5 26 25.5 OUTLET HEAD 31m 34m 25 24.5 24 0 0.5 1 1.5 Element Flow (l/s) 2 2.5
Reasons for Passive Flow Modulation Spring & Diaphragm Fixed Throttle Pilot (Variable Throttle) Vent to air (closed) Main Valve Diaphragm Pilot Isolation Valve Model modulates K up and down to achieve set outlet pressure User defined K(min) represents maximum valve open position
Representing CAPACITY limitations The importance of K: Software Default Default Wide-Open (Manufacturer) k 18.6 Calibrated (80% Open) K - 31
Representing CAPACITY limitations Verifying Calibration of K Max recommended design opening 70% Observable deterioration starts at 70% opening (7l/s, k=50) Max opening achieved stabilises at 80% (k=31) Given this PRV is performing as expected Reasonable to assume that c. 80% maximum opening is typical this style of valve under pilot operation
Representing CAPACITY limitations Evidence of Upstream Pressure Profile in Downstream Profile: Indicates that this 80mm PRV is stuck at 64% open Evidence of Upstream Pressure Profile in Downstream Profile (shouldn t happen on functioning PRV) PRV has no (or very limited) capacity for modulation no evidence of control K is primary parameter for addressing capacity issues. Set K = 175 with a set pressure slightly above maximum observed to achieve calibration
SUMMARY Can Control Valves be Modelling Dynamically? PRVs are generally predictable in their behaviour if not ideal Software has the capacity to represent d/s head variation that can be linked to the influence of flow Most variations downstream of a functioning PRV can be related to the influence of flow (inconsistent with THV performance) Locked or seized PRVs can be modelled by selecting an appropriate K to represent the fixed valve opening position Capacity is primarily a function of valve geometry and can be satisfactorily estimated in K. Using no more than 80% open value for traditional style valve would be appropriate to limit risk associated with over predicting PRV capacity.
SUMMARY Can Control Valves be Modelling Dynamically? TO HAVE CONFIDENCE USING MODELS FOR: Resilience Analysis Incident Modelling Developer Services Growth Analysis Fire flow modelling Network Optimisation Capital Investment Solution Development Control Valves MUST be modelled dynamically