Practical Engineering — Recreating an Ancient Pump (with no moving parts)

Why this is in the vault

13-minute Grady Hillhouse build-along on the pulser pump — a centuries-old water-pumping device with no moving parts originally used at the Alhambra fortress in Granada, Spain (Renaissance through the 18th century, reconstructed by Spanish engineering professor Caserus in 1911). The pump uses falling water to entrain air bubbles, then uses those bubbles’ buoyancy to lift a smaller volume of water higher than its source. The vault keeps it for two reasons that aren’t “cool old engineering.” (1) The terminology-drift observation is the load-bearing transferable lesson — Grady documents that the device has at least three names in the literature (pulser pump, hydraulic air compressor, trompe + airlift composition) and the most recent Alhambra hydraulics paper doesn’t even name it, which is itself why the technology stayed lost. This is the engineering-domain twin of the vault’s own tag-discipline argument and adds a 4th source to CA-012 (Notation is the conceptual move): canonical terminology is what makes prior work re-discoverable. (2) The composition pattern (pulser pump = trompe + airlift pump) is the right design discipline for RDCO skills — every skill should be similarly decomposable into named sub-components so the parts can be re-composed. The “low efficiency, zero moving parts” trade-off is also a legitimate design point for RDCO autonomous-loop infrastructure where 3am skill failures are expensive and reliability beats speed.

Episode summary

13-minute Grady Hillhouse build-along on the pulser pump (a.k.a. air-lift pump driven by a trompe), a centuries-old water-pumping device with no moving parts originally used at the Alhambra fortress in Granada, Spain (Renaissance through the 18th century). The pump uses falling water to entrain air bubbles, then uses those bubbles’ buoyancy to lift a smaller volume of water higher than its source. Re-built in his garage from sponsor-provided acrylic. The deeper thesis: low-efficiency, high-reliability infrastructure with no electricity, no moving parts, and inherent corrosion-resistance is useful, forgotten, and under-named in the literature (the device is alternately called pulser pump, hydraulic air compressor, and unnamed-by-historians at the Alhambra). Closes with the same SendCutSend sponsor placement as the spillway video.

Key arguments / segments

[00:00:00] The Alhambra anchor. Granada’s medieval fortress complex, UNESCO site. The Alcasaba (the higher fortress on the western tip) was supplied by water using a pump with no moving parts through the Renaissance and possibly earlier.
[00:01:00] The 1764 priest’s description and the 1911 Caserus reconstruction. A priest observed the device in 1764 but couldn’t explain how it worked. Spanish engineering professor Caserus rebuilt it in 1911 from the priest’s description and presented results at a Granada scientific congress. Recent historians find the reconstruction plausible but not certain.
[00:02:00] The thesis: it’s clever, effective, and mostly unknown. “You can’t pick one up off the shelf at your local hardware store. At least not yet.” Frames the build as both a tribute to Caserus and a small intervention against forgotten engineering.
[00:02:30] Cool-because-purely-mechanical aside. Grady’s stated aesthetic: water + ingenuity + no electronics has something special. Prior builds in the same vein: trompe (water-powered air compressor), hydraulic ram pump (water-powered pump using check valves).
[00:03:30] Build setup with SendCutSend acrylic. Sponsor read embedded into the build narrative — “an entire idea from my head shipped to my door.” Demos the basin, downpipe, separator, riser, and discharge in turn.
[00:04:30] Step 1: the upper basin. Free water surface at the top — establishes the energy starting point. Equivalent to the end of an ancient canal feeding the system.
[00:05:00] Step 2: water + air entrainment via tee-fitting whirlpool. A tee fitting in the downpipe acts as a vent, drawing air into the falling stream. Caserus tried multiple methods; whirlpool drawdown worked best. Bubble-rich water descends rapidly.
[00:05:30] Step 3: air-water separation in the lower tank. Sloped top traps rising bubbles; water continues. The right-hand exit is now bubble-free water.
[00:06:00] At this midpoint the device is basically a trompe. Air is now under pressure (because of the riser back-pressure on the right). Trompes were used historically for smelting, powering tools, blowing-air applications. Modern interest: aerating mine-waste pools to accelerate bacterial cleanup of contamination.
[00:07:00] Step 4: the airlift discharge. Bubbles rise out of the separator into a narrow vertical pipe, dragging water upward between bubble pulses. Output: pulses (not continuous) of water above the inlet basin elevation. Water is being moved uphill with no moving parts.
[00:07:30] The airlift sub-pattern is itself ubiquitous. Used in wastewater plants, dredges, and (mechanism-equivalent) coffee percolators (steam bubbles lifting water to drip through grounds). Useful where impellers would get clogged.
[00:08:30] The two key efficiency caveats. (1) “All the water running through the pump does not actually get pumped” — the pumped fraction is “less than 5%” of the input flow rate by Grady’s eyeball estimate. (2) This is not a free-energy device — it’s converting falling-water energy into bubble-compression energy then into lifted-water energy, with substantial losses at each stage (analogous to a hydropower turbine, not a perpetual-motion machine).
[00:09:00] The decomposition: pulser pump = trompe + airlift pump. Two well-understood sub-systems composed in series.
[00:09:30] Two-phase flow is the unsolved-engineering bit. Air-water mixed flow is hard. Bubble size, downpipe diameter, mixing geometry, discharge pipe diameter all interact non-linearly. Lots of design knobs, sparse engineering literature.
[00:10:30] The terminology problem is itself a discovery problem. Grady found one paper in the Journal of Applied Thermal Engineering calling the device a “hydraulic air compressor.” A recent Alhambra hydraulic-techniques paper doesn’t even name the device. Most surviving info is “a few old YouTube videos and college projects.” Lack of standard terminology is itself why the technology stayed lost.
[00:11:30] The grad-student thesis pitch. Grady explicitly suggests pulser pumps would make a good thesis topic — there’s clear room for engineering improvement and clear potential applications (rural areas with abundant water but no electricity).
[00:11:45] The Caserus 1911 quote. “This arrangement will always have over the hydraulic ram the advantage of eliminating valves entirely since it contains no moving solid parts. Doing away with the ram stroke seems to remove any source of fatigue in the pipes and of course the very annoying noise that makes the ram inapplicable near living quarters.” (Quote slightly compressed.) The engineering pitch hasn’t changed in 114 years; the technology just got forgotten.
[00:12:30] Closing principle: revive the idea for new applications. “In cases where efficiency is not critical, reviving it could solve practical problems using a layout so simple that it is remarkable it has not become common knowledge after several centuries.”
[00:01:30 / embedded throughout] SendCutSend sponsorship integration. Sponsor is woven into the build narrative rather than appended at the end — same SendCutSend pitch as the spillway video.

Notable claims

[00:00:30] The Alhambra Alcasaba was water-supplied by a no-moving-parts pump through the Renaissance and possibly the medieval period until the late 18th century, when a higher canal replaced it.
[00:01:00] Spanish engineer Caserus reconstructed the device in 1911 from a 1764 priest’s description and presented results at a Granada scientific congress. Recent historians find the reconstruction plausible.
[00:08:30] Grady’s eyeball estimate: pumped flow is less than 5% of total flow rate through the pump. This is the headline efficiency penalty — you need a lot of falling water to lift a small amount higher.
[00:10:30] The technology has at least three names in the literature — pulser pump, trompe + airlift composition, hydraulic air compressor — and the Alhambra-specific paper doesn’t name it at all. Terminology drift is itself a barrier to rediscovery.
[00:11:00] Hydraulic ram pumps (the better-documented cousin) are popular in rural electrification-deficit communities. Pulser pumps occupy the same niche but are less noisy (no ram stroke, no valves, no fatigue cycles).
The credit prompt for the discovery is Primal Space’s video on Alhambra hydraulics (linked in the description); pairs with this Practical Engineering build as a two-channel collaboration on lost-engineering content.

Guests

None directly featured, but Primal Space (separate YouTube channel) is credited as the source that surfaced the topic — link in description. Caserus (1911) is the historical figure whose reconstruction the build re-traces.

Mapping against Ray Data Co

The terminology-drift point is the load-bearing RDCO lesson. A technology with three different names in the literature plus an unnamed mention in a recent paper is invisible to search. The vault has the same exposure: when the same concept appears under three different tag-spellings or three different filename-slugs, no future query surfaces all three. Direct corollary to CA-012 (Notation is the conceptual move — already in CANDIDATES.md). Grady’s frustration with the pulser pump literature is the engineering-domain twin of the vault’s tag-discipline justification. Worth strengthening CA-012 with this as a 4th supporting source — terminology drift in engineering literature is the same pattern as tag-drift in the vault, and the cost is identical (lost discoverability).
The composition pattern (pulser pump = trompe + airlift) is the right pattern for skill design. Grady’s decomposition isn’t novel — it just makes the design legible. Pulser pump = a trompe (water → pressurized air) chained with an airlift pump (pressurized air → lifted water). RDCO skills should be similarly composable and named-at-each-stage. The current /check-board skill is implicitly a (notion-fetcher → triage → fan-out-spawner) chain; making the sub-skills first-class would let them be re-composed (e.g., the triage sub-skill could be reused by /process-inbox).
The “useful + forgotten + under-named” failure mode applies to internal RDCO patterns. When something works for a year and then stops being top-of-mind, it gets re-invented next time the founder hits the same problem — costing time and creating drift. Worth a periodic /vault-health flag for patterns documented but never re-cited — patterns that risk becoming “the pulser pump” of RDCO ops.
Low-efficiency-but-zero-maintenance is a legitimate design point for RDCO infrastructure choices. Pulser pumps are <5% efficient but have zero moving parts. RDCO faces the same trade-off in MCP server choice: a “less efficient” approach (e.g., always re-fetch from API rather than cache) may have zero state-management overhead and fewer failure modes. Worth being explicit when low-efficiency / high-reliability is the right trade-off — especially for the autonomous loop where 3am skill failures are expensive.
The grad-student-thesis pitch maps to RDCO’s /curiosity skill. Grady explicitly invites grad students to research pulser pumps — high-leverage research with clear application. The /curiosity skill should be biased toward this kind of question: well-defined, well-bounded, applicable, under-researched. Worth filing “look for engineering domains with terminology drift / lost knowledge” as a /curiosity prompt-pattern.
This video is the pair-piece to the Spillway video for CA-016 (Layered-defense). Both videos in the same backfill, both same author, both essentially about engineering for predictable failure modes. The Spillway video is the layered-defense exemplar (fuse plug as the second-layer EMA-equivalent). This video is the zero-moving-parts inherent reliability exemplar — different angle on the same overall philosophy. Strengthens the concept-page draft.

Open follow-ups

Add the terminology-drift point to CA-012 (Notation is the conceptual move) as a 4th source. Already-ripe candidate; this strengthens it. ~5 min vault edit.
Audit the vault for “pulser pump” patterns — documented but never re-cited. Run /vault-health with a “patterns documented in concepts/ that have zero inbound wikilinks from any post-creation doc” filter. Surface candidates for either re-promotion or archival. ~30 min.
Add “low-efficiency / high-reliability” as an explicit design-decision row in the SKILL.md template. Every new skill spec should declare which axis it’s optimizing. The pulser-pump example is the spec. ~5 min.
Sanity Check angle: “The pulser pump problem — when terminology drift hides good ideas.” Open with the Alhambra device (visceral, photogenic, undeniable). Pivot to the data-engineering equivalent: data marts that get re-built every 18 months because nobody remembers the previous mart existed under a different name. Land on the editorial discipline: canonical terminology is what makes prior work re-discoverable. ~1500 words. Pairs naturally with a CA-012 concept page.
Add Primal Space to the YouTube watch list (Tier 2). Cross-channel collaboration validates the channel’s quality. Video on Alhambra hydraulics was credited as the surfacer. ~10 min channel-id lookup.

Sponsorship

The video carries a SendCutSend placement — same sponsor as the spillway video, integrated into the build narrative at ~3:30 rather than appended at the end. Per RDCO bias-flagging discipline:

The technical content (Alhambra history, Caserus 1911 reconstruction, trompe + airlift decomposition, two-phase-flow caveats, terminology-drift observation) is editorial — drawn from public history, engineering literature, and the producer’s domain expertise.
The SendCutSend placement is paid sponsorship; the acrylic parts for this build were provided. Should be discounted as marketing, not as a vetted product recommendation.
Cross-promotion of Primal Space is not paid (creator-to-creator credit + link), but worth flagging as an audience-share arrangement common between technical-explainer YouTube channels.

~/rdco-vault/06-reference/transcripts/2026-04-20-practical-engineering-ancient-pump-no-moving-parts-transcript.md — full transcript
~/rdco-vault/06-reference/2026-04-20-practical-engineering-spillway-failed-on-purpose — paired Practical Engineering piece on engineered-failure-mode hydraulic infrastructure
~/rdco-vault/06-reference/concepts/CANDIDATES.md — adds 4th source to CA-012 (Notation is the conceptual move) via the terminology-drift angle