Wrong fixes and why they fail
Tank 1 starts at 27 ft with a 40 ft overflow elevation, leaving 13 ft of storage. Its reported inflow is 7,500 gpm, versus 1,100 gpm into Tank 2. The number that matters is net storage flow: pump inflow minus flow through the connecting pipe and any unreported outlet. A 30 s calculation interval can represent that balance, but a small interval cannot correct uncertain pump flow, incorrect elevations, or a bad pipe-flow curve.
| Attempt or symptom | Why it fails | Engineering correction |
|---|---|---|
| Use 7,500 gpm and 1,100 gpm as constant historical facts | A pump's operating flow can change as tank level raises discharge head. The stated values may be nominal, indicated, or measured at only one operating point. | Reconstruct each inflow from calibrated flow history or from the pump curve and time-varying system head. |
| Assign the unexplained Tank 2 rise to the 16 in connection | Transfer is driven by hydraulic-grade difference, not by Tank 1 inflow. The initial liquid-level difference is only 3 ft. | Calculate transfer from the free-surface elevations, pipe resistance, valve state, and fitting losses. |
| Apply one unsigned polynomial for every level difference | Flow reverses when the hydraulic-grade difference changes sign. A squared term alone does not encode direction and extrapolation can produce an unphysical result. | Solve flow magnitude from head loss, then apply the sign of . |
| Refine the time step until the story appears to match | Numerical precision is not input accuracy. Unknown outlets, valve movements, or changing pump rates dominate the result. | Use sensitivity cases and compare predicted levels against independent timestamps. |
Hydraulic mechanism
The two vessels exchange water only when their hydraulic grades differ. For vented tanks containing water, the driving head is the difference between free-surface elevations referenced to one datum. Equal liquid depths represent equal hydraulic grades only when the tank bases share the same elevation. A riser, elevated connection, closed or throttled valve, trapped gas, check valve, or different base elevation changes the effective system.
For a bottom-to-bottom connection with both tanks at the same base elevation, Tank 1 initially has 3 ft more head. Flow therefore starts from Tank 1 toward Tank 2. The pipe flow reduces Tank 1's accumulation and increases Tank 2's accumulation by the same amount. It cannot create or remove water from the combined system.
The reported connection is a 16 in carbon-steel pipe approximately 300 ft long. Pipe diameter, length, roughness, fittings, valve position, entrance losses, and exit losses determine how much transfer a given head difference produces. As levels move, both transfer magnitude and possibly direction move with them.
Quantities and deciding measurements
| Quantity | Reported value or limit | Where to read or verify it |
|---|---|---|
| Tank 1 initial level | 27 ft | Calibrated level history and local gauge |
| Tank 2 initial level | 24 ft | Calibrated level history and local gauge |
| Overflow elevation | 40 ft | Tank drawing and surveyed overflow point |
| Tank 1 reported inflow | 7,500 gpm | Flowmeter history, pump operating data, and valve history |
| Tank 2 reported inflow | 1,100 gpm | Flowmeter history, pump operating data, and valve history |
| Interconnecting line | 16 in, 300 ft, carbon steel | Piping drawing, field walkdown, and valve lineup |
| Calculation interval | 30 s | Model configuration; compare with a halved interval |
| Tank cross-sectional area | Not stated directly | Certified tank dimensions or strapping table |
| Tank-base elevations | Must share a common datum | Survey and civil drawings |
| Unreported tank outlets | Unknown | Lineup, flow history, and operating log |
Two closed-connection timing checks imply the tank geometry used in earlier calculations. Tank 1 rising 13 ft at 7,500 gpm for gives about 84,231 gal/ft. Tank 2 rising 16 ft at 1,100 gpm for gives about 84,563 gal/ft. These differ by about 0.4% and correspond to a cylindrical diameter near 120 ft. Treat that diameter as a reverse-calculated check, not a replacement for the tank drawing.
Governing mass and energy balances
For constant-area tanks, write both balances with one signed transfer flow:
A1 dh1/dt = Q1,in - Q12 - Q1,out
A2 dh2/dt = Q2,in + Q12 - Q2,out
Q12 is positive from Tank 1 to Tank 2. Adding the equations cancels the intertank flow. That cancellation supplies the strongest mass-balance check:
d(V1 + V2)/dt = Q1,in + Q2,in - Q1,out - Q2,out
Calculate the connection from Darcy-Weisbach with minor losses:
|H1 - H2| = [f(L/D) + sum(K)] v^2/(2g)
Q12 = sign(H1 - H2) Ap v
Re = rho v D / mu
The friction factor f depends on Reynolds number and relative roughness. A noniterative correlation valid across laminar and turbulent regimes can simplify each step, but its inputs still need the actual inside diameter, roughness, and fluid properties. Near zero head difference, preserve the sign explicitly and return zero flow when the grades are equal.
One proposed fitted relationship was Q_cfs = 0.0283 dH^2 + 0.0363 dH, with dH in feet. At the initial 3 ft difference it gives 0.3642 ft3/s, or approximately 163 gpm. Used with that fit, initial net flows are about 7,337 gpm into Tank 1 storage and 1,263 gpm into Tank 2 storage. The fit requires validation against the Darcy-Weisbach system curve over the entire modeled head range; its positive quadratic term makes extrapolation especially sensitive at larger head differences.
Time-step reconstruction procedure
Establish one elevation datum. Convert both level readings to free-surface elevations and confirm the tank bases, pipe connection elevations, overflow elevation, and any high points.
Obtain each tank's area-versus-level or volume-versus-level table. Equal diameter and height support equal geometry, but certified dimensions decide the volume calculation.
Build time-dependent boundary conditions for pump flow, outlet flow, and valve position. If only pump information exists, calculate the operating point from the pump curve and system head at each level rather than holding a nominal flow constant.
At the beginning of each 30 s interval, calculate . Solve the pipe equation for signed
Q12, including straight-pipe and minor losses.Calculate net flow for each tank. Convert gallons to volume units compatible with tank area, then update level with
Delta h = Qnet Delta t/Aor use the tank strapping table for nonconstant area.Repeat until Tank 1 crosses 40 ft. Interpolate within the final interval between the last level below 40 ft and the first level at or above it.
Repeat with half the time interval. A materially different overflow time indicates that the original interval is too large or the flow solver is discontinuous near reversal.
With the connecting valve closed and no outlets, the model should reproduce approximately 2 h 26 min for Tank 1 and for Tank 2 using the geometry implied by those benchmarks. This isolates integration and unit errors before pipe transfer is introduced.
Flow-history and scenario corrections
Constant inflow is the largest investigative assumption. A centrifugal pump operating against a rising liquid level normally moves along its pump curve unless a flow-control system holds the commanded rate. Flow indication, pump speed, discharge pressure, valve position, and start-stop status separate a controlled constant-flow case from a changing operating point.
Reported scenarios for Tank 2 contain values that require different hidden assumptions. Raising Tank 2 from 24 ft to 30 ft in was associated with both 3,600 gpm and 5,922 gpm. A further estimate used 11,280 gpm to rise from 30 ft to 40 ft in one hour. Separate gross pump inflow, pipe transfer, outlet flow, and net storage flow before using any of these numbers.
An illustrative coupled-tank case used about 7,000 gpm into Tank 1 and 8,500 gpm into Tank 2 to place both overflows near . That case demonstrates the required direction of correction: Tank 2 needs substantially more water than the reported 1,100 gpm if both tanks approached overflow together. It does not establish the actual pump flows.
Verification against the event timeline
| Check | Passing result | Failure indicated |
|---|---|---|
| Combined mass balance | Total modeled storage change equals total external inflow minus external outflow | Unit error, omitted stream, or incorrect level-to-volume conversion |
| Transfer direction | Flow follows the sign of the hydraulic-grade difference | Sign convention or elevation-datum error |
| Closed-valve boundary case | Tank 1 reaches overflow near 2 h 26 min under the stated constant inflow and implied geometry |
Area, time, or gallon conversion error |
| Time-step convergence | Halving 30 s changes the crossing time only negligibly for the investigation's required resolution | Step too large or unstable pipe-flow calculation |
| Observed intermediate levels | Calculated levels pass through independent gauge readings at their timestamps | Changing pump rate, missing outlet, valve movement, or instrument error |
| Pipe-curve range | Every modeled head lies within the validated hydraulic calculation or fitted range | Unsupported extrapolation |
Corroborate the operating account with a band of timelines rather than one false-precision timestamp: a nominal case, credible minimum and maximum pump-flow cases, alternative valve states, and level-instrument uncertainty. The account is hydraulically possible when its claimed overflow time and intermediate observations fall within cases that also close the combined mass balance.
Frequently asked questions
How do I calculate the Tank 1 overflow time?
At each 30 s step, calculate signed pipe transfer, subtract it and any outlet from Tank 1 inflow, and convert the net volume to level rise using the tank area or strapping table. Stop at the interpolated crossing of 40 ft.
How do I check the model before adding pipe flow?
Close the connection mathematically and set outlets to zero. With the stated constant flows and the geometry implied by the timing checks, Tank 1 should take about 2 h 26 min and Tank 2 about to reach 40 ft.
How do I decide whether the 16-inch pipe explains Tank 2's rise?
Calculate transfer from free-surface elevation difference using the verified pipe and valve resistance, then compare it with Tank 2's required net storage flow. At the initial 3 ft difference, the stated fitted curve predicts only about 163 gpm; validate that fit before applying it at later heads.
How do I know when to stop the reconstruction and escalate?
Stop when no combination within measured flow, level, valve, and geometry uncertainty can satisfy both the event timestamps and combined mass balance. Escalate unresolved flowmeter, pump-curve, level-instrument, or hydraulic-model discrepancies through the equipment manufacturer's official support channel. Provide trend exports, calibration records, drawings, valve lineup, pump operating data, and the failing balance case.