pump-system-integration - SKILL.md Agent Skill

name: pump-system-integration description: "Design complete pump systems including piping, controls, and parallel/series configurations" category: thinking domain: mechanical complexity: advanced dependencies: [numpy, networkx]

Pump System Integration

Overview

Comprehensive pump system design integrating pumps with piping networks, controls, and multiple pump configurations for industrial applications.

System Design Workflow

1. System Requirements Analysis

Define Operating Parameters:

Required flow rate range (Q_min, Q_normal, Q_max)
Total head requirements (static + dynamic)
Fluid properties (density, viscosity, temperature)
Operating conditions (continuous, intermittent, variable)
Redundancy requirements (N+1, duty/standby)
Environmental constraints

System Classification:

Open vs. closed loop systems
Constant vs. variable volume systems
Batch vs. continuous process
Critical vs. non-critical service

Performance Metrics:

Efficiency targets
Energy consumption limits
Lifecycle cost constraints
Reliability requirements (MTBF)

2. System Curve Development

Static Head Components:

Elevation difference: H_static = Z_discharge - Z_suction
Tank pressure differences: ΔP_tank
Process pressure requirements

Dynamic Head Components:

Friction losses (Darcy-Weisbach): h_f = f × (L/D) × (v²/2g)
Minor losses (fittings, valves): h_m = K × (v²/2g)
Velocity head changes
Equipment pressure drops

System Curve Equation:

H_system = H_static + K_system × Q²
where K_system accounts for all friction losses

Variable System Curves:

Multiple operating scenarios
Seasonal variations
Worst-case design points
Safety margins (10-15% typical)

3. Pump Selection and Matching

Duty Point Selection:

Locate operating point: intersection of pump curve and system curve
Verify pump efficiency at duty point (within 85-90% of BEP)
Check NPSH available vs. required (margin ≥ 0.5-1.0 m)
Ensure operation within preferred operating region (POR)

Pump Sizing Criteria:

Normal flow: 80-110% of BEP flow
Avoid operation below 70% or above 120% of BEP
Consider future capacity requirements
Evaluate turndown ratio for variable flow

Multiple Pump Options:

Single large pump vs. multiple smaller pumps
Fixed speed vs. variable speed
Trade-offs: capital cost, efficiency, redundancy, control

4. Piping Layout and Sizing

Velocity Limits:

Suction piping: 0.6-1.5 m/s (minimize losses, prevent cavitation)
Discharge piping: 1.5-3.0 m/s (balance cost vs. friction)
High-velocity concerns: erosion, noise, water hammer

Piping Best Practices:

Eccentric reducers on suction (flat side up to prevent air pockets)
Straight pipe runs: 5-10D upstream, 2-3D downstream
Minimize elbows and fittings near pump
Support piping to prevent stress on pump nozzles
Isolation valves for maintenance
Drain and vent provisions

Pipe Sizing:

D = √(4Q / πv)
Select standard pipe size (DN, NPS)
Calculate actual velocity and pressure drop

Pressure Drop Calculation:

Friction factor (Moody diagram or Colebrook equation)
Reynolds number: Re = ρvD/μ
Relative roughness: ε/D
Total equivalent length: L_eq = L_pipe + ΣL_fittings

5. Parallel vs Series Pump Configurations

Parallel Pumps Configuration:

Applications:

Variable flow requirements
Redundancy (N+1 reliability)
High flow, moderate head systems
Turndown capability

Combined Characteristics:

At same head: Q_total = Q_1 + Q_2 + ... + Q_n
Flow distributes based on individual pump curves

Combined H-Q Curve:

For identical pumps: double flow at each head
For different pumps: add flows at constant heads
Operating point shifts right (higher flow)

Flow Distribution:

Pumps operate at same discharge pressure
Flow divides according to individual resistances
Check valve prevents backflow through idle pumps
Balance piping resistances to equalize flow

Control Sequencing:

Lead/lag selection (rotate for wear equalization)
Staging based on flow demand
Minimum runtime between starts
Automatic switchover on failure

Parallel Pump Issues:

Unstable operation if pumps have drooping curves
One pump may dominate if curves differ significantly
System resistance must be sufficient (avoid runout)
Need check valves to prevent reverse flow

Series Pumps Configuration:

Applications:

High head, moderate flow requirements
Boosting pressure in long pipelines
Multi-stage processes
Overcoming elevation changes

Head Addition:

At same flow: H_total = H_1 + H_2 + ... + H_n
Heads add at each flow rate

Combined H-Q Curve:

For identical pumps: double head at each flow
For different pumps: add heads at constant flows
Operating point shifts up (higher head)

Staging Considerations:

Pump order based on head contribution
Inter-stage pressure limitations
NPSH available for downstream pumps
Temperature rise considerations

Series Pump Issues:

Downstream pumps must handle increased pressure
Check shaft power requirements at all stages
Inter-stage piping may need pressure class upgrade
Fluid temperature increases with each stage

6. Control Strategies

Variable Frequency Drive (VFD) Control:

Affinity Laws for Speed Variation:

Q_2/Q_1 = N_2/N_1
H_2/H_1 = (N_2/N_1)²
P_2/P_1 = (N_2/N_1)³

Advantages:

Energy savings at reduced flow (cubic relationship)
Smooth flow control
Soft start reduces mechanical stress
Eliminates throttling losses

Disadvantages:

Higher capital cost
Potential harmonic issues
Minimum speed limitations (cooling, lubrication)
Not suitable for all applications (steep system curves)

VFD Energy Savings:

Power savings at reduced flow:
P_ratio = (Q_ratio)³ for VFD
P_ratio = varies for throttling (less efficient)

Throttling Control (Discharge Valve):

Method:

Increase system resistance by closing valve
Shifts system curve upward
Operating point moves left on pump curve

Characteristics:

Simple and low cost
Energy inefficient (creates artificial resistance)
Suitable for occasional flow adjustment
Better for flat system curves

Bypass Control:

Recirculates excess flow back to suction
Maintains minimum flow through pump
Prevents overheating at low flows
Very energy inefficient

On/Off Control:

Simplest method
Suitable for batch processes
Pressure or level switches
Consider cycle frequency limits

Advanced Control:

PID control loops (pressure, flow, level)
Cascade control systems
Predictive control algorithms
System optimization (minimize total energy)

7. Transient Analysis (Water Hammer)

Water Hammer Causes:

Rapid valve closure
Pump startup/shutdown
Check valve slam
Air pocket collapse

Joukowsky Equation (Pressure Surge):

ΔP = ρ × a × Δv
where:
a = wave speed = √(K/ρ) / √(1 + (K/E)(D/t))
K = bulk modulus of fluid
E = elastic modulus of pipe
Δv = velocity change

Critical Closure Time:

T_critical = 2L/a
If closure time < T_critical: maximum pressure surge
If closure time > T_critical: reduced pressure surge

Mitigation Strategies:

Slow valve closure (increase closure time)
Surge tanks or accumulators
Air chambers (cushioning effect)
Surge relief valves
Pump bypass systems
Controlled pump coastdown

Analysis Requirements:

Transient simulation software (AFT Impulse, WANDA, Hammer)
Pipe material properties
Valve closure characteristics
Pump inertia and performance curves
Allowable pressure limits (pipe rating ×1.5)

Parallel Pumps Analysis

Combined H-Q Curves

For Identical Pumps:

# At each head value:
Q_combined = n × Q_single
where n = number of pumps running

# Operating point:
Solve: H_pump(Q_combined/n) = H_system(Q_combined)

For Different Pumps:

# Create combined curve:
for H in head_range:
    Q_total = Q_pump1(H) + Q_pump2(H) + ...

# Find intersection with system curve

Graphical Method:

Plot individual pump curves
At each head, add flows horizontally
Plot system curve
Intersection = operating point
Individual flow = total flow × (individual resistance ratio)

Flow Distribution

Equal Resistance Distribution:

Identical pumps, symmetric piping: equal flow
Q_each = Q_total / n

Unequal Resistance:

Flow distributes inversely with resistance:
Q_1/Q_2 = √(R_2/R_1)

For n pumps with different heads at same flow:
All pumps operate at same discharge pressure P_d
Q_i determined by individual pump curve at P_d

Balancing Techniques:

Balance valves in each pump discharge
Symmetric piping layout (equal length, fittings)
Flow meters for monitoring
Pressure taps for verification

Control Sequencing

Demand-Based Staging:

Flow ranges for n-pump operation:
1 pump: Q_min to Q_1max
2 pumps: Q_1max to Q_2max
3 pumps: Q_2max to Q_3max

Start additional pump when:
- Current flow > upper limit
- Current pump > 90% capacity
- Pressure drops below setpoint

Stop pump when:
- Flow < lower limit with n pumps
- Redundant capacity available
- Pressure exceeds setpoint

Lead/Lag Rotation:

Equalize runtime across pumps
Alternate lead pump daily/weekly
Automatic rotation algorithm
Manual override capability

Sequencing Logic:

IF system_flow > Q_setpoint_high AND pumps_running < pumps_available:
    START next_pump
    DELAY minimum_time

IF system_flow < Q_setpoint_low AND pumps_running > 1:
    STOP lagging_pump
    DELAY minimum_time

ROTATE lead_pump EVERY rotation_period

Series Pumps Analysis

Head Addition

Identical Pumps in Series:

# At each flow value:
H_combined = n × H_single
where n = number of pumps in series

# Operating point:
Solve: n × H_pump(Q) = H_system(Q)

Different Pumps in Series:

# Create combined curve:
for Q in flow_range:
    H_total = H_pump1(Q) + H_pump2(Q) + ...

# Find intersection with system curve

Graphical Method:

Plot individual pump curves
At each flow, add heads vertically
Plot system curve
Intersection = operating point
Each pump delivers same flow, different heads

Staging Considerations

Pump Order:

Generally high-head pumps downstream
Consider NPSH requirements
Minimize inter-stage pressure
Temperature rise accumulation

Inter-stage Pressure:

P_interstage = P_suction + H_1 × ρ × g
Must not exceed:
- Downstream pump casing rating
- Piping pressure class
- Seal pressure limits

NPSH Cascade:

NPSH_available for pump n = P_discharge(n-1) - P_vapor + H_static - H_losses
Verify: NPSH_a > NPSH_r + margin for all stages

Temperature Rise:

ΔT = (H × g × (1 - η)) / (c_p × η)
where:
H = head per pump
η = pump efficiency
c_p = specific heat

Total rise = Σ(ΔT_i) for all pumps
Check: T_final < fluid limits, seal limits

Staging Strategies:

All pumps always on (simple, reliable)
Sequential staging (variable head applications)
Bypass first stage at low demand
VFD on first stage, fixed speed on boosters

Piping Network Analysis

Network Modeling

Node-Based Analysis:

Nodes: junctions, tanks, pumps
Links: pipes, valves, fittings
Continuity at each node: ΣQ = 0
Energy balance around loops: ΣH = 0

Hardy-Cross Method:

Iterative solution for complex networks:
1. Assume flow distribution
2. Calculate head loss in each pipe: h = K × Q²
3. Calculate loop corrections: ΔQ = -Σh / (2ΣK|Q|)
4. Update flows: Q_new = Q_old + ΔQ
5. Repeat until convergence

Matrix Methods:

[A] × [H] = [Q]
where:
[A] = network connectivity matrix
[H] = nodal heads
[Q] = nodal demands

Solve using linear algebra

Common Network Configurations

Branching System:

Tree structure, single path to each point
Simple analysis: sequential calculation
No loops, no redundancy

Looped System:

Multiple paths between points
Hardy-Cross or matrix solution required
Redundancy, better pressure distribution

Distributed Demand:

Withdrawal along pipe length
Approximate with multiple point loads
Use equivalent length methods

Pressure Analysis

Minimum Pressure Requirements:

Process equipment inlet pressures
End-user pressure requirements
Elevation effects
Future expansion margin

Maximum Pressure Limitations:

Pipe pressure ratings (ANSI class)
Equipment pressure limits
Valve and fitting ratings
Safety factors (typically 1.5-2.0)

Critical Points:

Highest elevation (low pressure risk)
Furthest from pump (highest resistance)
High demand nodes
Branch takeoffs

System Optimization

Energy Optimization

Minimize Total Energy Consumption:

E_total = Σ(P_pump × t) + E_losses
where:
P_pump = (ρ × g × Q × H) / η
E_losses = pumping, piping, control losses

Optimize:
- Pump efficiency (operate near BEP)
- Pipe sizing (balance capital vs. energy)
- Control strategy (VFD vs. throttling)
- System curve reduction (improve layout)

Lifecycle Cost Analysis:

LCC = C_capital + C_energy + C_maintenance + C_downtime

C_energy = (P_pump × hours × $/kWh × years)
Typically dominant for continuously operating systems

Optimization Variables:

Pump selection (size, type, speed)
Pipe diameter (velocity vs. friction)
Number of pumps (parallel/series)
Control method (VFD, throttle, on/off)

Reliability Optimization

Redundancy Configurations:

N+1: One standby for N duty pumps
N+2: Two standbys (critical applications)
2×100%: Two pumps, each capable of full duty
3×50%: Three pumps, any two handle full load

Reliability Calculations:

Series reliability: R_series = R_1 × R_2 × ... × R_n
Parallel reliability: R_parallel = 1 - (1-R_1)(1-R_2)...(1-R_n)

Availability = MTBF / (MTBF + MTTR)

Maintenance Strategies:

Predictive maintenance (vibration, performance)
Preventive maintenance schedules
Condition monitoring
Spare parts inventory

Design Optimization Process

Define Objective Function:
- Minimize energy cost
- Minimize capital cost
- Minimize lifecycle cost
- Maximize reliability
Identify Constraints:
- Flow rate limits
- Pressure limits
- Budget constraints
- Space limitations
- Standardization requirements
Select Design Variables:
- Pump size and number
- Pipe diameters
- Control strategy
- Operating schedules
Optimization Methods:
- Parametric studies
- Gradient-based optimization
- Genetic algorithms
- Multi-objective optimization
Sensitivity Analysis:
- Vary key parameters
- Identify critical factors
- Assess robustness
- Evaluate risk

Practical Guidelines

System Design Checklist

System requirements clearly defined
System curve developed for all operating scenarios
Pump selection optimized for efficiency
NPSH margin verified (min 0.5-1.0 m)
Piping velocities within limits
Parallel/series configuration analyzed
Control strategy selected and sized
Water hammer analysis completed
Pressure ratings verified throughout system
Energy consumption calculated
Lifecycle cost evaluated
Redundancy requirements met
Maintenance accessibility verified
Instrumentation and monitoring specified

Common Design Mistakes

Undersized NPSH margin: Use adequate safety factor
Poor pump selection: Operating far from BEP
Excessive pipe velocities: Erosion and noise issues
Inadequate transient analysis: Water hammer damage
Improper parallel pump application: Flow instability
Ignoring system curve changes: Future conditions
Over-reliance on throttling: Energy waste
Insufficient redundancy: Reliability issues
Poor piping layout: Cavitation, air entrainment
Neglecting lifecycle costs: Focus only on capital cost

Performance Verification

Acceptance Testing:

Flow rate verification (±5% tolerance)
Head measurement at duty point
Power consumption check
Vibration and noise levels
NPSH margin confirmation

Monitoring Points:

Suction and discharge pressure
Flow rate
Power consumption
Vibration levels
Bearing temperature
Seal condition

Performance Trending:

Track efficiency over time
Detect degradation early
Schedule maintenance proactively
Optimize operating conditions

Examples

Example 1: Parallel Pump System Design

See network-model.py for complete implementation with:

Two pumps in parallel serving variable demand
Combined H-Q curve calculation
Operating point determination
Flow distribution analysis
Control sequencing logic

Example 2: Series Pump System Design

See network-model.py for complete implementation with:

Two-stage series configuration
Head addition calculation
Inter-stage pressure verification
NPSH cascade analysis
Temperature rise calculation

Example 3: Piping Network Analysis

See network-model.py for complete implementation with:

Hardy-Cross network solver
Multiple pump/demand configuration
Pressure distribution calculation
Optimization for pipe sizing

References

See reference.md for detailed standards and codes
Hydraulic Institute Standards (HI 9.6.3 - Pump System Design)
ASME B31.1/B31.3 Piping Codes
AFT Software - Transient Analysis
Crane TP-410 - Flow of Fluids

Notes

Always verify pump operation within POR (70-120% of BEP)
Consider future expansion in initial design
Energy costs typically dominate lifecycle costs
Reliability requirements drive redundancy decisions
System optimization requires balanced approach: capital, energy, maintenance