Late-Summer Reservoir Stratification: Choosing an Intake Depth When Every Port Costs You Something
A 6-MGD conventional surface water plant draws from a 1,800-acre stratified reservoir through a multi-port intake tower with ports at 12, 22, and 32 ft. In late August the thermocline sits at 18 ft, and the operator has been drawing from the 32-ft port since July to avoid an epilimnetic algal bloom. Over ten days: raw manganese 0.02 to 0.22 mg/L, iron 0.05 to 0.45 mg/L, dissolved oxygen at the intake 0.3 mg/L, raw pH 7.2 to 6.7, raw turbidity steady at 1.6 NTU, and hydrogen sulfide odor in the raw. Finished manganese has reached 0.07 mg/L; customers report black staining and a musty taste. Filter runs have shortened from 48 to 30 hours and chlorine demand has risen about 40%. The plant can feed potassium permanganate and powdered activated carbon, and can change intake depth at will.
1. Explain what is happening in the reservoir and why the intake depth that was the correct choice in July has become the wrong one. What data would you collect before moving the port?
Read the worked answer
Mechanism. Thermal stratification has isolated the hypolimnion below 18 ft. With no reaeration and continued biological oxygen demand from settled algae and sediment, dissolved oxygen at the 32-ft port has fallen to 0.3 mg/L. Under anoxic conditions manganese and iron are chemically reduced to soluble Mn2+ and Fe2+ (hence the rise from 0.02 to 0.22 mg/L Mn and 0.05 to 0.45 mg/L Fe), sulfate is reduced to sulfide (the rotten-egg odor), and decay-derived CO2 depresses pH (7.2 to 6.7).
Why this matters downstream, not in the reservoir. Soluble Mn2+ passes through coagulation and filters almost untouched. It is oxidized to colloidal MnO2 later, in the clearwell or in the distribution system, which is exactly why customers see black staining while the plant sees only 1.6 NTU of raw turbidity. The MnO2 solids also load the filters and the reduced species (Mn2+, Fe2+, sulfide, reduced organics) exert oxidant demand, so chlorine demand rises even though raw turbidity is flat.
Why July's choice is now wrong. In July the hypolimnion was still oxygenated and algal-free, so deep withdrawal was a clean win. By late August the deep water has become a reducing environment. The 12-ft port is oxygenated and low in Mn/Fe, but it carries the algae you were avoiding: expect higher turbidity, higher TOC and DBP precursor loading, and more filter solids. There is no free option, only a trade.
Data before moving the port. Vertical profiles at 3-ft intervals, weekly: temperature and DO (confirm the thermocline position and the anoxic boundary), pH, Mn, Fe, sulfide, ammonia, turbidity, and algal counts/identification. Also run bench tests on blended water (e.g., 2/3 epilimnion + 1/3 hypolimnion) for coagulant demand, settled turbidity, and chlorine demand. Move the port to a depth that is above the anoxic boundary and monitor Mn breakthrough, rather than assuming the shallowest port is best.
2. Finished manganese is 0.07 mg/L against a 0.05 mg/L secondary limit. Size a potassium permanganate feed and state the control points and failure modes you would watch.
Read the worked answer
Dose basis. The stoichiometry for oxidizing dissolved Mn2+ to MnO2 is about 1.92 mg KMnO4 per mg Mn2+ (2 MnO4- oxidize 3 Mn2+). At 0.22 mg/L raw Mn the theoretical requirement is 0.22 x 1.92 = about 0.42 mg/L. Real kinetics and competing demand (Fe2+, sulfide, organics) mean a practical dose of roughly 1.5 to 2 times theoretical, so about 0.6 to 0.9 mg/L, feed ahead of rapid mix with at least 3 to 5 minutes of contact and good mixing. Iron consumes permanganate too (about 0.94 mg KMnO4 per mg Fe2+), so if the raw Fe stays near 0.45 mg/L add roughly 0.4 mg/L of additional demand before you start trimming the Mn dose.
Control points.
- No persistent pink color at settled water (a visible permanganate residual means overfeed and will send color and MnO2 into the filters).
- Filtered Mn at or below the plant's internal target (well under 0.05 mg/L) and filtered turbidity stable, not rising from MnO2 solids.
- Settled and filtered turbidity trends; the MnO2 that forms is a fine colloid and may require a coagulant or polymer adjustment to capture.
- Permanganate feed rate paced to flow, with a knock-out or interlock should the feed lose pacing.
Failure modes. Overfeed causes pink water, elevated filter loading, and increased oxidant demand; underfeed passes soluble Mn through. PAC for the musty odor should be dosed at a separate point from permanganate, because PAC adsorbs odor compounds but also consumes oxidant and shortens filter runs. An alternative or supplement is raising pH above roughly 7.5 to 8.0 so chlorine can help oxidize Mn, but that trades against DBP formation and must be evaluated with the disinfection strategy.
3. The plant superintendent wants to know why chlorine demand climbed 40% when raw turbidity did not change. Answer, and say how you would verify it with a test.
Read the worked answer
Answer. Turbidity is a measure of light scattering, not of oxidant demand. The species driving the demand here are dissolved and largely invisible: Mn2+, Fe2+, sulfide, ammonia, and reduced organic matter leached from anoxic sediments. Each is a reduced chemical species that free chlorine will oxidize before any residual can be maintained. That is why a plant can see a flat 1.6 NTU raw turbidity and a sharply rising chlorine demand at the same time.
There is a second trap: if the plant is trying to oxidize Mn2+ with free chlorine alone at a depressed raw pH of about 6.7, the reaction is slow and incomplete. Chlorine appears to be "disappearing" into demand when part of the problem is that the oxidant is being applied at the wrong point and pH. Permanganate upstream, or pH adjustment before chlorination, will often collapse the apparent demand.
How to verify. Run a demand test on raw and settled water: measure immediate (15-minute) and long-term (e.g., 7-day at distribution temperature) chlorine demand, plus permanganate demand, on the current blend and on each intake port separately. Compare the 32-ft water against 12-ft water and against a blend. If the hypolimnetic sample shows high immediate demand and the epilimnetic sample does not, the reservoir profile is confirmed as the cause and the intake decision becomes a quantified trade rather than a guess. Repeat the test after permanganate oxidation to confirm the demand is being removed upstream rather than pushed into the clearwell.