I’ve spent 13 years designing industrial wastewater treatment plants. During that time, I’ve seen — and made — plenty of mistakes. Some were caught during construction. Others only surfaced after the plant was running, when operators called to ask why something wasn’t working the way the drawings said it would.
This article is about those mistakes. Real ones, from real projects. If you’re an engineer designing treatment plants, or a plant manager trying to figure out why your system isn’t performing, you’ll find something useful here.
1. Hydraulic Profile and Elevation Calculation Errors
The most common mistake I see is under-sizing pipes when the available hydraulic head is limited. Here’s an example from a coking wastewater project:
The design called for 30 m³/h of flow from a fluidized bed reactor to a secondary clarifier. Following standard practice, we sized the pipe at DN100 based on a 1 m/s velocity assumption. On paper, it looked fine.
In reality, the elevation difference between the two structures was smaller than anticipated. The DN100 pipe created too much resistance. Water backed up in the fluidized bed — you could see it overflowing at the seams.
The fix: We replaced the DN100 pipe with DN125. The problem disappeared immediately.
The lesson: When your available head is tight, don’t blindly follow the 1 m/s rule of thumb. Calculate the actual head loss through the entire pipe run — including fittings, bends, and the exit loss. If the numbers are close, go one size up. A slightly oversized pipe costs a little more upfront. An undersized one costs you every day the plant runs below capacity.
2. Process Selection for High-Chloride Pharmaceutical Wastewater
Last year, I worked on a pharmaceutical intermediate wastewater project. The chloride concentration was nearly 200,000 mg/L — essentially saltwater with organic compounds mixed in.
The Fenton Fiasco
Our initial design used the Fenton process. On paper, it made sense: Fenton is good at breaking down refractory organics, and the pharmaceutical intermediates in this wastewater were exactly the kind of persistent compounds Fenton is known for handling.
In practice, it was a disaster:
Chemical consumption went through the roof. The massive chloride concentration interfered with hydroxyl radical generation, so we had to dose far more reagent than stoichiometric calculations suggested.
Colloidal byproducts formed. These sticky colloids coated every surface they touched — glassware in the lab, pipes in the pilot unit, everything. You could scrape them off a beaker with your fingernail.
We abandoned Fenton and switched to a combination of pressurized biochemical treatment and catalytic oxidation. Performance improved immediately, and chemical costs dropped by an order of magnitude.
The Powdered Activated Carbon Trap
Before the equalization tank, we also tested powdered activated carbon (PAC) dosing combined with coagulation — hoping the PAC would adsorb COD and reduce the load on downstream treatment.
Two problems killed this approach:
The PAC saturated almost instantly. At these COD levels (several thousand mg/L), you’d need a truckload of fresh carbon every day to make a dent. Regeneration wasn’t practical, and the utilization rate was abysmal.
Sludge production exploded. PAC + coagulant + high organics = mountains of sludge. The sludge handling system we’d designed couldn’t keep up, and disposal costs alone made the economics unworkable.
The lesson: Before committing to an advanced oxidation or adsorption process, run bench-scale tests with the actual wastewater. Not synthetic water. Not “similar” water from another project. The real stuff. Two days of jar testing would have saved us weeks of pilot-scale frustration.
3. Front-End Pretreatment: Keep It Simple
In the early days of designing coking wastewater treatment trains, I often specified a front-end oil-separation system (dissolved air flotation or media filtration) followed by PAC coagulation as pretreatment.
The intent was sound: remove free oil and some COD before the biological stage. But there are two issues with over-engineering pretreatment:
If your downstream biological process can handle moderate oil loading, you’re adding capital cost and maintenance burden for marginal benefit.
Each pretreatment stage introduces its own failure modes. A clogged DAF unit or saturated filter isn’t protecting your bioprocess — it’s starving it, or sending slugs of untreated waste when bypassed.
I’ve since moved toward simpler front-end designs: coarse screening, adequate equalization volume (at least 12-24 hours of hydraulic retention), and temperature/pH correction. Let the biology do the heavy lifting.
4. Design for the Operator, Not Just the Drawing
Every engineer should spend one week shadowing plant operators before they’re allowed to finalize a layout drawing.
Here’s one that still makes me cringe: two pumps installed so close together that you couldn’t pull either one out. The clearance between them was less than the length of the pump casing. When one failed, maintenance had to disconnect the neighboring pump just to create enough room to extract the broken one.
This isn’t a rare edge case. Walk through any industrial park and you’ll find:
Valves positioned so high you need a ladder to reach them (and there’s no ladder)
Instrument panels facing walls
Gratings that can’t be lifted because a pipe runs 10 cm above them
Sampling points in locations no human can comfortably stand
The fix is simple: During the layout review, physically trace every maintenance task. How does an operator reach this valve? How does this pump get removed? Where does the rigging go? If you can’t answer those questions clearly, revise the layout.
5. Embedded Parts and the 2D-to-3D Leap
If I had a dollar for every embedded part that was installed in the wrong location, I’d have… well, a lot of dollars.
The root cause is almost always the same: 2D drawings can’t show clashes between embedded parts and pipe supports. A plan view shows the embed location. An elevation shows the pipe support. But nothing shows them intersecting until the contractor calls and says “these two things want to occupy the same space.”
Here are the problems I’ve encountered:
Anchor bolts embedded 200 mm off from where the equipment base actually landed
Pipe support base plates cast into concrete, then covered because the pipe route had to shift to avoid a ductbank
Wall penetrations that didn’t align with the equipment nozzle because someone rounded a dimension on one drawing but not the other
Switching to 3D design (P3D in my case) solved most of these. When you can orbit the model and see the pipe support bracket sitting right on top of your embed plate, the clash is obvious before concrete is poured. Collision detection in 3D modeling tools isn’t a nice-to-have — it’s a cost-avoidance mechanism that pays for itself on the first project.
Summary: Five Rules I Design By Now
When head is tight, oversize the pipe. The marginal cost is negligible compared to a hydraulic bottleneck.
Jar-test the real wastewater. No process selection decision should be made without actual bench-scale data from the specific waste stream.
Question every pretreatment stage. Each one adds capex, opex, and failure modes. If the biology can handle it, let it.
Walk through every maintenance task on the drawing. If an operator can’t comfortably reach it, see it, or remove it, redesign.
Design in 3D. The days of finding embedded part clashes during construction need to end.
These are lessons from real projects over 13 years in environmental engineering design. If you’ve encountered similar issues — or different ones — I’d love to hear about them in the comments.