World Energy Data · Field Development Insights

Region × development-type comparison (dry-tree vs wet-tree vs other) · Issue #776 · Offshore production-facility universe

worldenergydata ▸ Capabilities ▸ Insights ▸ Development type

Where dry trees give way to wet: the offshore development map

Deepwater is wet-tree country. Across 836 producing offshore facilities worldwide, the dry-tree share of trees collapses from 80% on the shallow shelf to 20% in ultra-deepwater (≥1 500 m) — surface well access does not survive the swim to the seabed.

836

Facilities

Regions

514

Dry-tree · 61%

266

Wet-tree · 32%

Other hull · 7%

Region × tree-type matrix

Every producing facility placed on the dry / wet / other trichotomy, by offshore region. Bars are facility counts; the trailing figure is the regional total.

Dry tree (surface trees on fixed jacket / compliant tower / TLP / spar)Wet tree (subsea trees: semisub-FPS / FPSO / FLNG / tieback)Other hull (FSO-FSU / MOPU / artificial island — no producing trees)

Region	Dry	Wet	Other	Total	Typed %
North Sea & NW Europe	155	69	11	235	95%
Asia-Pacific	141	72	22	235	91%
US Gulf of Mexico	64	23	1	88	99%
South America & Caribbean	30	50	1	81	99%
West Africa	31	40	6	77	92%
Mediterranean & N Africa	31	9	4	44	91%
Caspian & Russia	26	3	6	35	83%
Middle East	29	0	4	33	88%
Atlantic Canada & Arctic	7	0	1	8	88%

"Typed %" = share of the region's facilities that map to a dry- or wet-tree concept (i.e. not an FSO/FSU, MOPU or artificial island). South America & Caribbean and West Africa read wet-tree-heavy — those are the FPSO provinces; the Middle East and Caspian read all-dry — shallow, benign, jacket country.

The physical story: dry-tree share fades with water depth

Restricted to the 738 facilities with a recorded water depth. Bars are normalised across producing trees only (dry + wet; the few non-tree hulls are excluded), so the navy segment is the dry-tree share of trees. Read top-to-bottom, shallow to deep — the navy shrinks at every step.

Water-depth band	Dry	Wet	Other	Dry share of trees	n
0-150 m (shallow shelf)	350	86	44	80%	480
150-500 m (deep shelf)	53	48	4	52%	105
500-1500 m (deepwater)	40	68	0	37%	108
1500-3000 m (ultra-deepwater)	9	36	0	20%	45

Below 1 500 m, dry trees are 9 of 45 trees (20%); above it, 443 of 645 (69%). The nine deep dry-tree facilities are the TLPs and spars that push surface access to its engineering limit.

Who builds what: country leaders by tree type

Dry-tree leaders

UK	82
US	64
Norway	46
Malaysia	33
China	25
Australia	18

Wet-tree leaders

Brazil	49
UK	37
Norway	31
Australia	29
US	23
Angola	15

Other-hull leaders

Thailand	7
Nigeria	5
UK	5
Norway	5
Malaysia	5
Kazakhstan	4

Concept mix (as-built host type)

fixed_jacket465fpso152(unmapped/other)56subsea_tieback54semisub_fps52tlp25spar20flng8compliant_tower4

Coverage & honesty

This universe is the curated production_facilities.csv (836 standalone facilities). Two limits worth stating plainly:

Subsea tiebacks are under-counted. A tieback is not a standalone facility — it hangs off a host — so only 266 wet-tree hulls appear here, of which just the 54 that the source lists as their own line are tiebacks. Where we do have block-level ground truth (the GoM SubseaIQ–BSEE crosswalk, 75 typed fields), subsea tiebacks are 10 of them — the true wet-tree footprint is larger than a facility census shows.
The typed/untyped split is honest per region. 56 of 836 facilities (7%) are non-tree hulls (FSO/FSU, MOPU, artificial island) and are held out of the dry-vs-wet ratios rather than forced into one bucket. 98 facilities lack a recorded water depth and are excluded from the depth chart only.

Sources: data/modules/offshore_assets/curated/production_facilities.csv (836 rows, HOST_TYPE → concept → tree_type) · subseaiq_bsee_block_crosswalk.csv (115 matched GoM rows, supplementary subsea-tieback prevalence only). Trichotomy: dry = {fixed_jacket, compliant_tower, tlp, spar, nui} (imported verbatim from worldenergydata.field_development.recommendation._DRY_TREE); wet = {semisub_fps, fpso, flng, subsea_tieback, subsea_to_shore}; other = unmapped HOST_TYPE + null. Method: build_portfolio() imports cleanly but covers only the 115-field GoM matched set, so the all-region matrix is built directly from the facility census with the mapping above. Build: scripts/field_development/build_region_devtype_comparison.py → reports/field_development/region_devtype_matrix.csv. Counts are facilities, not fields or wells; every figure on this page is computed from the source rows (flag-don't-fake).