Theory explainer
Muskingum Routing Problems — FE/PE Civil Exam Worked Examples
A flood does not arrive downstream the way it left upstream. As the wave travels down a river reach it spreads out: the peak gets lower and it arrives later. Routing is the calculation that turns the hydrograph at the top of a reach into the hydrograph at the bottom, and the Muskingum method is the exam’s way of doing it — one storage equation, three coefficients, and a table you step through one row at a time. Learn the coefficient arithmetic and the time-step check, and every Muskingum problem below is a few lines.
Last reviewed: 2026-11-08. Every worked number below was independently re-solved; steps are shown in full.
What routing is: channel routing versus reservoir routing
Routing answers one question: given the hydrograph entering a reach, what hydrograph leaves it? There are two families, and the exam tests the difference between them. Reservoir routing (level-pool or modified Puls routing) applies where the water surface stays level — a reservoir behind a dam. Storage then depends on the outflow alone, because the pool surface is horizontal at every instant, and the routing runs off a storage–outflow curve. It is worked on the Flood Frequency and Reservoir Routing page. Channel routing applies where the water surface is sloped — a river reach. Storage then depends on both the inflow and the outflow, because the wedge of water stored in the reach is bigger on the rising limb, when inflow exceeds outflow, than on the falling limb, when outflow exceeds inflow. The Muskingum method is the exam’s channel-routing method: it is hydrologic routing, meaning it works with storage and continuity alone and never solves the momentum equation, which is what keeps it fast enough to do by hand in an exam. The price of that simplicity is calibration — K and x come from observed floods, not from channel geometry, unless you use the Muskingum–Cunge variant (see below).
The storage equation: prism storage and wedge storage
Picture a reach of river with a flood wave moving through it. At any instant the water stored in the reach has two parts. The prism storage is the volume that would be there if the flow were steady at the outflow rate: Sprism = K O, where K is a time constant with units of time. The wedge storage is the extra volume piled up while the inflow runs ahead of the outflow: Swedge = Kx (I − O). Add them and the storage equation appears: S = K O + Kx (I − O) = K [xI + (1 − x)O]. The two parameters carry the physics. K is approximately the travel time of the flood-wave peak through the reach — a long reach has a large K, a short reach a small one — and in the absence of data it is estimated that way. x is the weighting factor: it says how much of the storage follows the inflow. For a natural stream x lies between 0 and 0.3, with 0.2 the usual default; a steep, narrow channel has x near the top of the range, a wide flat one near the bottom. At x = 0 the wedge disappears, storage depends on the outflow alone, and the method collapses to the reservoir case — a linear reservoir. At x = 0.5 the storage follows the average of inflow and outflow and the wave is translated downstream with no attenuation at all, which is why the exam never asks for x above 0.5: it is the physical limit of the model. The storage S comes out in discharge × time — cfs·hr in US units, m³/s·hr in SI — so converting it to a volume needs a factor of 3,600 seconds per hour (or 86,400 per day); that conversion is a standing trap.
The routing equation: three coefficients and one check
The routing equation comes from writing the continuity equation over one step — the change in storage equals the average inflow minus the average outflow over Δt — and substituting the storage equation at the start and the end of the step. The algebra gives O2 = C1I2 + C2I1 + C3O1, with the coefficients C1 = (Δt − 2Kx)/(2K(1 − x) + Δt), C2 = (Δt + 2Kx)/(2K(1 − x) + Δt), and C3 = (2K(1 − x) − Δt)/(2K(1 − x) + Δt). Three facts about these coefficients do most of the exam work. First, they always sum to 1.0 — add the numerators and you get the denominator — so check the sum before you route anything; if it is not 1.0, a coefficient is wrong. Second, all three must be positive for a valid solution, which forces the routing interval into the window 2Kx ≤ Δt ≤ 2K(1 − x): a step longer than 2K(1 − x) makes C3 negative, a step shorter than 2Kx makes C1 negative, and either one produces a meaningless outflow — usually negative or oscillating. Third, the routing is a table, not a single calculation: O2 from one step becomes O1 for the next, and you step through the whole inflow hydrograph row by row. Worked Problem 1 does the coefficient arithmetic and two steps; Worked Problem 2 runs a full eight-step hydrograph and shows the peak attenuation and lag the method exists to produce.
Estimating K and x from observed floods
When a reach has gauged inflow and outflow hydrographs from a past flood, K and x are estimated by calibration — the loop method. First build the storage at each time step from the continuity equation: the change in storage over a step is the average inflow minus the average outflow, times Δt, accumulated from a datum. Then, for a trial value of x, compute the weighted discharge xI + (1 − x)O at each step and plot storage against it. The storage equation says S = K times that weighted discharge, so the correct x is the one that makes the plot closest to a single straight line — the wrong x leaves a loop, because the rising limb and the falling limb trace different paths — and the slope of that straight line is K. Worked Problem 4 runs the method on an observed flood with x = 0.2, and the two point pairs both give K = 10 hours, which is how you know the trial x is the right one. In the absence of observed data the exam accepts the defaults: K from the travel time of the peak through the reach, and x = 0.2 for a natural stream. The Muskingum–Cunge variant, which the PE exam names at depth, replaces calibration with channel properties: K = Δx/ck and X = ½(1 − Q/(B ck S0 Δx)), where Δx is the reach length, ck the kinematic wave celerity, B the channel top width, and S0 the bed slope. It is the same routing equation with physically based parameters — know the two formulas and the fact that they come from the channel rather than from a past flood, and that is the depth the exam asks for.
Worked problem 1 Routing coefficients and two steps (FE pace, US units)
Given: a river reach with K = 2.3 hr and x = 0.15. The routing interval is Δt = 1 hr. The inflow hydrograph is I1 = 93 cfs at hour 1, I2 = 137 cfs at hour 2, and I3 = 208 cfs at hour 3, with the outflow at hour 1 equal to O1 = 85 cfs. Find the outflow at hours 2 and 3.
Solution:
- Denominator, common to all three coefficients: 2K(1 − x) + Δt = 2 × 2.3 × 0.85 + 1 = 3.91 + 1 = 4.91.
- C1 = (Δt − 2Kx)/4.91 = (1 − 2 × 2.3 × 0.15)/4.91 = (1 − 0.69)/4.91 = 0.31/4.91 = 0.0631.
- C2 = (Δt + 2Kx)/4.91 = (1 + 0.69)/4.91 = 1.69/4.91 = 0.3442.
- C3 = (2K(1 − x) − Δt)/4.91 = (3.91 − 1)/4.91 = 2.91/4.91 = 0.5927. Check: 0.0631 + 0.3442 + 0.5927 = 1.0000, and all three are positive, so the step is valid.
- Outflow at hour 2: O2 = C1I2 + C2I1 + C3O1 = 0.0631 × 137 + 0.3442 × 93 + 0.5927 × 85 = 8.64 + 32.01 + 50.38 = 91.0 cfs.
- Outflow at hour 3: O3 = C1I3 + C2I2 + C3O2 = 0.0631 × 208 + 0.3442 × 137 + 0.5927 × 91.0 = 13.12 + 47.16 + 53.95 = 114 cfs.
Answer: O2 ≈ 91.0 cfs; O3 ≈ 114 cfs. Trap note: the coefficient pairing is the whole problem — C1 multiplies the new inflow I2, C2 the old inflow I1, and C3 the old outflow O1. Swapping I1 and I2 is the commonest slip, and the sum-to-1.0 check will not catch it, because the coefficients are right either way. The second trap is forgetting that O2 becomes O1 for the next step — step 6 uses the 91.0 cfs just computed, not the 85 cfs from hour 1.
Worked problem 2 A full hydrograph: attenuation and lag (FE pace, US units)
Given: a reach with K = 1.0 hr and x = 0.2, routed at Δt = 1 hr. The inflow hydrograph (cfs) at hours 0 to 7 is 100, 300, 600, 900, 700, 400, 200, 100, starting from steady flow so O0 = I0 = 100 cfs. Route the flood and find the outflow peak, how much it is attenuated, and how much it lags the inflow peak.
Solution:
- Denominator: 2K(1 − x) + Δt = 2 × 1.0 × 0.8 + 1 = 2.6. C1 = (1 − 0.4)/2.6 = 0.2308; C2 = (1 + 0.4)/2.6 = 0.5385; C3 = (1.6 − 1)/2.6 = 0.2308. Check: 0.2308 + 0.5385 + 0.2308 = 1.0001, and all three are positive.
- Step through the table, Oj+1 = 0.2308 Ij+1 + 0.5385 Ij + 0.2308 Oj:
For example, the hour-4 row: O4 = 0.2308 × 700 + 0.5385 × 900 + 0.2308 × 608 = 161.6 + 484.7 + 140.3 = 786 cfs.Hour Inflow (cfs) Outflow (cfs) 0 100 100 1 300 146 2 600 334 3 900 608 4 700 786 5 400 651 6 200 412 7 100 226 - Inflow peak: 900 cfs at hour 3. Outflow peak: 786 cfs at hour 4. Attenuation = 900 − 786 = 114 cfs; lag = 1 hour.
Answer: outflow peak ≈ 786 cfs at hour 4 — 114 cfs lower than the inflow peak and 1 hour later. Trap note: the outflow is still rising at hour 3 while the inflow is already falling — that is the lag, and it is the point of the exercise. An answer that puts the outflow peak at hour 3, or at 900 cfs, has skipped the routing entirely. Also note the outflow is still 226 cfs at hour 7, long after the inflow has returned to baseflow: the reach is still draining its stored wedge, which is the volume-conservation check — the area under the outflow hydrograph equals the area under the inflow hydrograph, minus nothing, because Muskingum routing conserves volume.
Worked problem 3 The time-step validity check (PE pace, US units)
Given: a reach with K = 6 hr and x = 0.3. (a) Is a routing interval of Δt = 12 hr valid? (b) If not, find the valid range of Δt and route one step with Δt = 6 hr, given I1 = 200 cfs, I2 = 350 cfs, and O1 = 180 cfs.
Solution:
- (a) Check Δt = 12 hr: the validity window is 2Kx ≤ Δt ≤ 2K(1 − x) = 2 × 6 × 0.3 = 3.6 hr to 2 × 6 × 0.7 = 8.4 hr. A 12-hour step is outside it. The coefficients confirm it: denominator = 2 × 6 × 0.7 + 12 = 20.4, and C3 = (8.4 − 12)/20.4 = −3.6/20.4 = −0.177 — negative, so the step is invalid and any outflow it produces is meaningless.
- (b) Route with Δt = 6 hr: denominator = 8.4 + 6 = 14.4. C1 = (6 − 3.6)/14.4 = 2.4/14.4 = 0.1667; C2 = (6 + 3.6)/14.4 = 9.6/14.4 = 0.6667; C3 = (8.4 − 6)/14.4 = 2.4/14.4 = 0.1667. Check: 0.1667 + 0.6667 + 0.1667 = 1.0001, and all three are positive.
- O2 = 0.1667 × 350 + 0.6667 × 200 + 0.1667 × 180 = 58.3 + 133.3 + 30.0 = 222 cfs.
Answer: Δt = 12 hr is invalid (C3 = −0.177); the valid range is 3.6–8.4 hr, and with Δt = 6 hr the routed outflow is ≈ 222 cfs. Trap note: a negative coefficient is not a negative flow and it is not a rounding error — it is the method telling you the step is wrong. The exam’s favourite version of this problem hands you a Δt that is too long and asks for the outflow: the correct answer is that the routing is invalid, or the re-routed value at a valid step. Never just run the arithmetic and report the number the negative coefficient produces.
Worked problem 4 Estimating K and x from an observed flood (PE pace, US units)
Given: the observed inflow and outflow hydrographs for a river reach, at 6-hour intervals, with storage computed from the continuity equation starting from a datum of 2,000 cfs·hr at t = 0:
| Time (hr) | Inflow (cfs) | Outflow (cfs) | Storage (cfs·hr) |
|---|---|---|---|
| 0 | 200 | 200 | 2,000 |
| 6 | 500 | 227 | 2,818 |
| 12 | 800 | 403 | 4,826 |
| 18 | 600 | 602 | 6,012 |
| 24 | 400 | 583 | 5,460 |
Using a trial value of x = 0.2, estimate K for this reach and check whether x = 0.2 is the best fit.
Solution:
- For each time, compute the weighted discharge xI + (1 − x)O with x = 0.2: at t = 0, 0.2 × 200 + 0.8 × 200 = 200 cfs; at t = 6, 0.2 × 500 + 0.8 × 227 = 100 + 181.8 = 282 cfs; at t = 12, 0.2 × 800 + 0.8 × 403 = 160 + 322.6 = 483 cfs; at t = 18, 0.2 × 600 + 0.8 × 602 = 120 + 481.2 = 601 cfs; at t = 24, 0.2 × 400 + 0.8 × 583 = 80 + 466.0 = 546 cfs.
- The storage equation says S = K × (weighted discharge), so K is the slope of S against the weighted discharge. Between t = 0 and t = 6: K = (2,818 − 2,000)/(282 − 200) = 818/82 = 10.0 hr.
- Check between t = 6 and t = 12: K = (4,826 − 2,818)/(483 − 282) = 2,008/201 = 10.0 hr. The same slope from both point pairs means the storage points fall on a single straight line for x = 0.2 — the loop has closed, so x = 0.2 is the best fit and K = 10 hr.
Answer: K ≈ 10 hr with x = 0.2 — the two independent point pairs give the same slope, confirming the fit. Trap note: the check is the point of the problem. A single pair of points always gives some K for any trial x; what makes x = 0.2 the answer is that a second, independent pair gives the same K. With the wrong trial x the two slopes would differ, and the S-versus-weighted-discharge plot would show a loop instead of a line. The second trap is the storage units: the table is in cfs·hr, so K comes out in hours directly — converting storage to cubic feet first (multiply by 3,600) and then dividing by cfs gives seconds, and mixing the two gives a K off by a factor of 3,600.
Worked problem 5 One routing step (SI units)
Given: a reach with K = 3 hr and x = 0.25, routed at Δt = 2 hr. Inflows are I1 = 40 m³/s and I2 = 90 m³/s, with the outflow at the start of the step O1 = 35 m³/s. Find O2.
Solution:
- Check the step first: 2Kx = 2 × 3 × 0.25 = 1.5 hr and 2K(1 − x) = 2 × 3 × 0.75 = 4.5 hr, so Δt = 2 hr is inside the valid window.
- Denominator: 2K(1 − x) + Δt = 4.5 + 2 = 6.5. C1 = (2 − 1.5)/6.5 = 0.5/6.5 = 0.0769; C2 = (2 + 1.5)/6.5 = 3.5/6.5 = 0.5385; C3 = (4.5 − 2)/6.5 = 2.5/6.5 = 0.3846. Check: 0.0769 + 0.5385 + 0.3846 = 1.0000.
- O2 = 0.0769 × 90 + 0.5385 × 40 + 0.3846 × 35 = 6.92 + 21.54 + 13.46 = 41.9 m³/s.
Answer: O2 ≈ 41.9 m³/s. Trap note: the outflow (41.9) is below both the new inflow (90) and the average inflow — that is the reach storing water on the rising limb, and it is correct. A candidate who expects the outflow to sit between I1 and I2 has confused routing with averaging. The units are also the point of this problem: K in hours and Δt in hours cancel in the coefficients, so the same formulas work in SI and US units unchanged — what changes is only the discharge unit carried through.
Frequently asked questions
What is the Muskingum method?
The Muskingum method is a hydrologic (lumped) method for routing a flood hydrograph through a river reach. It combines the continuity equation with a linear storage equation, S = K[xI + (1 - x)O], and steps the outflow forward in time with the routing equation O2 = C1 x I2 + C2 x I1 + C3 x O1. The result is an outflow hydrograph whose peak is lower than the inflow peak and arrives later — the attenuation and lag a real flood wave shows as it travels downstream.
What do K and x mean in the Muskingum method?
K is the storage time constant of the reach, with units of time; it is approximately the travel time of the flood-wave peak through the reach. x is a dimensionless weighting factor between 0 and 0.5 that says how much the storage depends on the inflow versus the outflow: x = 0 means storage depends on outflow alone (a reservoir), and natural streams usually fall in the range 0 to 0.3, with 0.2 a common default. Worked Problem 4 shows how both are estimated from observed inflow and outflow hydrographs.
How do you calculate the Muskingum routing coefficients?
From K, x and the routing interval dt: C1 = (dt - 2Kx) / (2K(1 - x) + dt), C2 = (dt + 2Kx) / (2K(1 - x) + dt), and C3 = (2K(1 - x) - dt) / (2K(1 - x) + dt). Then O2 = C1 x I2 + C2 x I1 + C3 x O1. The three coefficients always sum to 1.0 — check that before routing, because a wrong coefficient is the commonest arithmetic slip. Worked Problem 1 runs the full calculation.
How do you estimate K and x from observed flood data?
Build the storage at each time step from the continuity equation, then plot storage S against the weighted discharge xI + (1 - x)O for a trial value of x. The x that makes the plot closest to a single straight line is the best fit, and the slope of that line is K. Worked Problem 4 runs the method on an observed flood: with x = 0.2 the storage points fall on a straight line of slope 10 hours, so K = 10 hours and x = 0.2.
What happens if the routing time step is too long or too short?
All three routing coefficients must be positive for a valid solution, which requires 2Kx <= dt <= 2K(1 - x). A time step longer than 2K(1 - x) makes C3 negative; a step shorter than 2Kx makes C1 negative. Either way the routing is invalid and the outflow it produces is meaningless — usually a negative or oscillating discharge. Worked Problem 3 shows a 12-hour step failing the check for K = 6 hours and x = 0.3, then re-routes the same flood with a valid 6-hour step.
What is the difference between Muskingum routing and reservoir (level-pool) routing?
Muskingum routing is channel routing: storage in a river reach depends on both the inflow and the outflow, because the wedge of water stored on the rising limb exceeds the wedge on the falling limb. Level-pool (modified Puls) routing is reservoir routing: the pool surface is level, so storage depends on the outflow alone. Muskingum with x = 0 is the reservoir limit — storage becomes S = K x O, the wedge disappears, and the two methods agree. The reservoir form is worked on the Flood Frequency and Reservoir Routing page.
Free FE/PE flashcards + study tips, by email
Join the Water Resources Hub email list and get the free 40 water-resources flashcards (PDF + Anki) straight away, then study tips and practice problems by email. Email address only — no spam, unsubscribe any time.