I've been handling wastewater treatment equipment sourcing orders for nine years, mostly for industrial and mining sites. My mistake log has six entries on it. Together they cost roughly $140,000 in wasted budget, and four of the six were sourcing-path decisions, not equipment failures.

That's the part nobody warns you about. The equipment rarely breaks the project. The way you bought it does.

The Two Paths, and the Standard I'm Using

There are two ways to buy a wastewater treatment plant's worth of equipment, and I've done both badly.

Path A — Packaged: you hand one OEM a package scope. Pretreatment, equalization, chemical dosing, clarifier or DAF, sludge handling, controls. They deliver a system that hits an effluent number.

Path B — Unbundled: you buy the pieces separately. Tanks from a fabricator, pumps from a pump house, blowers from a blower supplier, membrane or media skids from a specialist, controls from an integrator, and you stitch the interfaces together on site.

Everyone has an opinion about which is better. Most of those opinions come from someone selling one of the two. So I'm comparing them on six dimensions I actually track — price, lead time, accountability, spares, change orders, and commissioning — with a straight conclusion on each. One of those conclusions is the opposite of what I assumed for the first three years.

Dimension 1: Sticker Price vs. Installed Cost

Unbundled wins on the quote. It usually does, and by a lot — I've seen equipment-only quotes come in 18–25% under a packaged equivalent.

The gap closes once you add what the quote left out: interconnecting piping, structural steel, cabling, engineering hours to reconcile three vendors' conflicting assumptions, and the integrator's fee for making someone else's skid talk to someone else's PLC.

Concrete case. A 250 m³/h coal washing water system, 2021. Unbundled equipment came to $318,000. The packaged quote was $402,000 — 26% higher. I pushed for unbundled. Final installed cost was $511,000 against $470,000 for the package. My 26% saving turned into a 9% loss, plus about 200 hours of my team's time that never showed up on any spreadsheet.

Equipment price is a number. Installed cost is the decision.

Conclusion: Unbundled wins the quote and loses the budget in most greenfield builds. If you can't produce an interface matrix before you issue the PO, the packaged price is probably the honest one.

Dimension 2: Lead Time — Where I Had It Backwards

This is the one that surprised me, and it's the reason I stopped trusting my own instincts on packaged systems.

I assumed a single vendor meant a shorter schedule. One PO, one production queue, no coordination. In 2019 I ordered a packaged system for a plant expansion assuming exactly that. It took 22 weeks. A comparable unbundled scope I'd run the year before took 14.

The reason is embarrassingly simple once you see it. An OEM builds your package in sequence — tank first, then internals, then skid assembly, then controls, then FAT. Unbundled, those five things happen in parallel across five shops. Parallel beats sequential almost every time, as long as you can manage five relationships instead of one.

I couldn't, back then. That's why I defaulted to packaged. But I'd also assumed packaged was faster, which was never true — I just hadn't checked.

Conclusion: Unbundled usually wins on lead time, not because suppliers are faster, but because the work runs in parallel. Packaged wins on lead time only when your own coordination capacity is the bottleneck. If you have no one to chase five vendors, 22 weeks is a fair price for not having to.

Dimension 3: Single-Point Accountability

Packaged wins this, and it isn't close.

When the effluent misses the limit at commissioning, a packaged scope gives you one party to call. An unbundled scope gives you a conference call where five suppliers explain, correctly and in detail, that their equipment met spec and the problem is the interface.

2020, and this one still stings. We bought a clarifier from one vendor and sized the feed pumps off a flow assumption from a different vendor's equalization tank design. Nobody owned the number between them. The clarifier went hydraulically overloaded, solids carryover wrecked the downstream filters, and it took nine weeks and $41,000 to sort out. The clarifier was fine. The pumps were fine. The gap between them was not, and the gap was mine.

Worth noting for the mining side: effluent limits for coal and ore mining categories are set federally under 40 CFR Part 434 and Part 440 respectively, and state permits often layer on tighter numeric limits. Check the current e-CFR text and your actual discharge permit before anyone writes a performance figure into a contract — the subparts have been amended over the years, and a limit you remember from a 2015 project may not be the limit you're held to. When one party has signed up to that number in writing, with consequences attached, you've bought something real. When five parties have each signed up to their own slice of it, you've bought a coordination problem with a compliance risk attached.

Conclusion: If the performance obligation is enforceable and the effluent limit is tight, packaged. Pay the premium. It's cheaper than being the integrator.

Dimension 4: Spares and the Ten-Year Tail

Unbundled wins, clearly and boringly.

Off-the-shelf pumps, blowers, and instruments can be second-sourced. A proprietary packaged system's control panel cannot.

I learned this when a single I/O board failed on a packaged system in 2022. The panel program was locked, the board was sole-source, and the replacement was $6,800 and three weeks — three weeks of a plant running at half capacity because nobody had asked, at PO stage, what happens when this board dies.

Two questions fix this, whichever path you take. Who owns the source code, and what's the spare parts list for year 5 through year 10? Ask before award. Vendors answer honestly when it's still a competitive bid.

Conclusion: Unbundled, unless you've contractually secured open access to the control system and a documented spares list. That's a paperwork fix, not a reason to abandon packaged systems — but it only works if you do it before the PO.

Dimension 5: Change Orders

This one is genuinely a wash, but for opposite reasons, and knowing which reason applies to you matters.

Packaged change orders are expensive and predictable. The scope is rigid, the vendor prices variations at a markup, and you know roughly what a change will cost within a day of asking.

Unbundled change orders look cheap individually and are expensive in aggregate. Each supplier's variation is small. But every change ripples through the interface, so you re-coordinate piping, wiring, and controls — three line items you never see on one invoice. My 2021 project had eleven small variations totaling $9,400 in supplier change orders and roughly $31,000 in downstream rework and engineering. Nobody ever approved the $31,000. It just appeared.

Conclusion: If your scope is likely to change, packaged if you can afford the markup, unbundled if you have engineering capacity to absorb the ripple. Choose based on your bench, not the change order rate.

Dimension 6: Commissioning

Unbundled gets you running fast and then leaves you holding the finger-pointing. Packaged includes commissioning — but read what "commissioning" means in the quote.

In my experience, "commissioning included" has meant everything from three days of a technician watching your crew work, to a full performance test with written acceptance criteria. The gap between those two is worth tens of thousands and a month of schedule.

Specify it. Days on site, deliverables, who supplies the test water, who pays for the crane, and the acceptance protocol. For pumps, the Hydraulic Institute's acceptance test standards define grades and tolerances — pick one and name it, or you'll spend a week arguing about 3% at the test stand. For motors, confirm the duty rating in writing (IEC 60034-1, S1 continuous), because a motor rated for intermittent duty will fail in a wastewater application eventually. And if your degasser or pressure filter runs above 15 psig, that vessel needs an ASME Section VIII Div. 1 stamp — which determines who can even fabricate it, and that affects your vendor list before it affects your price.

Conclusion: Packaged, with an acceptance protocol written into the purchase order. Unbundled commissioning works only if you have someone on site who can be the integrator of record.

So Which One Do You Pick

Four dimensions to packaged, two to unbundled, and one wash. That doesn't mean packaged wins, because the dimensions aren't equal weight — a $41,000 interface failure outweighs a $9,000 price delta every time.

Go packaged when the effluent limit is tight and enforceable, the site has no engineering bench of its own, footprint is constrained, and you can get the control system access and spares list written into the contract. Tight compliance plus no in-house engineers is the classic packaged case. Don't apologize for the premium.

Go unbundled when you have process and mechanical engineers on staff, the site is remote with long spares logistics, capacity will expand in phases over several years, and you can enforce an interface matrix. Mining sites with staged expansion fit this well — you want the pumps and blowers to be locally serviceable in year eight.

Or do what I do now: package the process core — biological, membrane, or chemical treatment where the performance risk lives — and unbundle the balance of plant. Tanks, pumps, piping, electrical, and structural steel are commodity scopes with commodity risk. The treatment step is where the compliance exposure is, and that's where a single accountable party is worth paying for.

One page, twelve lines, before any PO goes out. Who sizes what. Who supplies which flange. What the acceptance test is. Who pays for the crane. Twelve lines I didn't have in 2017, and every one of the six entries in my mistake log traces back to a line that wasn't there.

Five minutes of interface checking beats five weeks of conference calls. I've paid for that lesson twice. You don't have to.