Harvest System Registry

FHOPS ships with a typed registry describing commonly-used harvest systems. Each system lists the ordered jobs, required machine roles, and precedence relationships; solver components and data contracts rely on this metadata to enforce sequencing constraints.

Data Model

Harvest systems live under fhops.scheduling.systems.models and are represented by two dataclasses:

SystemJob

Defined by name, machine_role, and a list of prerequisite job names.

HarvestSystem

Holds system_id, ordered jobs, and optional environment / notes metadata.

The helper fhops.scheduling.systems.models.default_system_registry() returns the built-in collection summarised below.

Default Systems

System ID	Environment	Machine Roles (ordered)	Notes
ground_fb_skid	ground-based	feller-buncher → grapple skidder → roadside processor → loader	Mechanised felling followed by grapple skidding to roadside processing and loading.
ground_fb_loader_liveheel	ground-based	feller-buncher → grapple skidder → roadside processor → loader	Same chain as ground_fb_skid but the loader job is pinned to the Barko 450 live-heel preset (TN-46) for interior operations.
ground_hand_shovel	ground-based	hand faller → shovel logger → roadside processor → loader	Hand falling with shovel logging as primary transport.
ground_fb_shovel	ground-based	feller-buncher → shovel logger → roadside processor → loader	Mix of mechanised felling and shovel logging before roadside processing.
ground_salvage_grapple	ground-based salvage	feller-buncher → grapple skidder → roadside processor → loader	ADV1N5 burned-timber salvage defaults (buck/sort fire damage, double-ring debarkers, charcoal controls) threaded into the grapple-skidder chain.
ctl	cut-to-length	single-grip harvester → forwarder → loader	Harvester processes at stump and forwarder hauls shortwood to trucks.
steep_tethered	steep-slope mechanised	tethered harvester → tethered shovel/skidder → roadside processor → loader	Winch-assist equipment across the full sequence.
cable_standing	cable-standing skyline	hand/mech faller → skyline yarder → landing processor/hand buck → loader	Standing skyline with chokers feeding landing processing.
cable_standing_tr125_single	cable-standing skyline	hand/mech faller → standing skyline (TR125 single span) → landing processor → loader	Skylead C40 single-span defaults (1.6 m³ payload, 25 m lateral). Use when you want the TR125 helper and grapple_yarder_skyleadc40 cost role.
cable_standing_tr125_strip	cable-standing skyline (intermediate supports)	hand/mech faller → standing skyline (TR125 multi-span) → landing processor → loader	Strip-cut defaults pulling the TR119 strip-cut treatment multiplier and setting lateral distance to 40 m (TN-258 warning auto-fires).
cable_salvage_grapple	cable-running salvage	hand/mech faller → grapple yarder → landing processor → loader	ADV1N5 salvage workflows for steep burned slopes (parallel bunching, grapple yarding, charcoal-aware processing) bundled into a cable-running preset.
cable_running	cable-running skyline	hand/mech faller → grapple yarder → landing processor/hand buck → loader	Grapple yarder variant with landing finishing.
cable_running_adv5n28_clearcut	cable-running skyline	hand/mech faller → long-span skyline → landing processor/hand buck → loader	ADV5N28 clearcut conversion (Madill 071 + Pow’-R Block replacing helicopter yarding).
cable_running_adv5n28_shelterwood	cable-running skyline	hand/mech faller → long-span skyline → landing processor/hand buck → loader	ADV5N28 irregular shelterwood conversion threading riparian corridors with full suspension.
cable_running_fncy12	cable-running skyline (intermediate supports)	hand/mech faller → skyline yarder → landing processor/hand buck → loader	Thunderbird TMY45 + Mini-Mak II preset (FNCY12/TN258). Selecting this system preloads the fncy12-tmy45 helper so TN-258 lateral-distance warnings and Cat D8/Timberjack support ratios flow automatically. Cost reference: see the Skyline cost matrix below (grapple_yarder_tmy45).
cable_micro_ecologger	cable-short-span skyline	hand faller → RMS Ecologger skyline → hand buck/processor → loader	TN173 NB case study (2.9 logs/turn, 0.34 m³ pieces, four-person crew) for compact skyline corridors.
cable_micro_gabriel	cable-short-span skyline	hand faller → Gabriel truck yarder → hand buck/processor → loader	TN173 Stephenville prototype skid-pan yarder (0.16 m³ stems, road-portable) with chokers and manual processing.
cable_micro_christie	cable-short-span skyline	hand faller → Christie tower yarder → hand buck/processor → loader	TN173 Christie hot-yarding configuration (two-person crew, 0.49 m³ pieces) for patch cuts.
cable_micro_teletransporteur	cable-short-span skyline	hand faller → Télétransporteur carriage → hand buck/processor → loader	TN173 Télétransporteur downhill hot-yarding (0.21 m³ pieces, self-propelled carriage) for riparian buffers.
cable_micro_timbermaster	cable-short-span skyline	hand faller → Smith Timbermaster skyline → hand buck/processor → loader	TN173 Timbermaster downhill skyline (0.54 m³ pieces, trailer tower) for small shortwood corridors.
cable_micro_hi_skid	cable-short-span skyline	hand faller → Hi-Skid yard/load/haul (direct to dump)	FNG73 Hi-Skid short-yard truck (100 m reach, 12 m³ deck); use --model hi-skid and optionally --hi-skid-include-haul when this system is selected. Cost reference: skyline_hi_skid entry in the Skyline cost matrix.
cable_partial_tr127_block1	cable-standing skyline partial cut	hand/mech faller → standing skyline (TR127 Block 1) → landing processor → loader	Provides the dual lateral-distance defaults (15 m + 6 m) and now applies the SR109 patch-cut penalty (volume ×0.846, cost ×1.10) via the built-in partial-cut profile.
cable_partial_tr127_block5	cable-standing skyline partial cut	hand/mech faller → standing skyline (TR127 Block 5) → landing processor → loader	Sets 16 m lateral, 3 logs/turn, 1.6 m³ payload, and wires in the SR109 green-tree penalty (volume ×0.715, cost ×1.10) so telemetry logs the added cost/range limits automatically.
helicopter	helicopter	hand faller → helicopter longline → hand buck/processor → loader/water	Helicopter longline with on-landing processing or direct-to-water transfer.

Using the Registry

• Scenario contract: blocks may specify harvest_system_id; the scenario definition can embed a harvest_systems map to override or extend the defaults (see fhops.scenario.contract.models).

• Synthetic generator: fhops.scenario.synthetic.generate_with_systems() assigns systems round-robin to blocks and creates matching machine role inventories.

• Solvers: sequencing constraints in the MIP and heuristics consume the registry to enforce job ordering and machine-role feasibility. Violations surface through KPI outputs and CLI summaries.

Scenario Cross-Links

The sample datasets shipped in examples/ reference these system IDs directly. Use the following map when you need to trace a block back to the registry table above:

Scenario	Where to inspect	Notes
examples/synthetic/{small,medium,large}	data/blocks.csv (harvest_system_id column) and README.md	Every synthetic bundle assigns an explicit system ID per block. The README describes the terrain/prescription mix; look up the IDs in the table above to recover the associated job/role chain.
examples/med42 / examples/large84	README.md and data/blocks.csv (terrain/prescription hints)	These scenarios omit harvest_system_id in order to let researchers swap systems during experiments. The README files call out the typical ground/cable combinations used in regression tests; add a column to data/blocks.csv if you need deterministic system assignments.
examples/tiny7	README + CLI quickstart	The toy dataset keeps a single ground-based workflow; tie blocks to ground_fb_skid or ground_hand_shovel when testing sequencing logic.

For any other dataset, run rg --no-heading 'harvest_system_id' path/to/blocks.csv to confirm whether the IDs are specified. When missing, edit blocks.csv or include a harvest_systems block inside scenario.yaml so future contributors know which registry entries to reference.

Extending Systems

To define custom systems, construct HarvestSystem instances and supply them via the scenario contract. When adding optional or parallel tasks, document the intended precedence explicitly and consider updating this reference alongside any new data files.

Preset Usage Cheat Sheet

Forwarder Productivity

Use fhops dataset estimate-productivity --machine-role forwarder and select a model from fhops.productivity.forwarder_bc.ForwarderBCModel. The CLI echoes the same arguments as fhops.productivity.forwarder_bc.estimate_forwarder_productivity_bc(); the docstring there lists all parameters and return fields. Pick a preset based on the stand context:

--forwarder-model	When to use	Key CLI options	API reference
ghaffariyan-small / ghaffariyan-large	Plantation or thinning work with 14–20 t forwarders where distance and slope dominate cycle time.	--extraction-distance, --slope-class or --slope-factor.	fhops.productivity.forwarder_bc.estimate_forwarder_productivity_bc()
adv6n10-shortwood	Boreal multi-product sorting (ADV6N10) where payload, log length, and travel speed are known.	--payload-per-trip, --mean-log-length, --trail-length, --travel-speed, --products-per-trail.	fhops.productivity.forwarder_bc.ADV6N10ShortwoodResult
adv1n12-shortwood	Valmet 646 regression driven primarily by extraction distance (Advantage 1 No. 12).	--extraction-distance only; slope handled via the preset.	fhops.productivity.forwarder_bc.ADV1N12ShortwoodResult
eriksson-final / eriksson-thinning	Swedish follow-up study with payload-driven productivity—use when you have stem size and load capacity data.	--mean-extraction-distance, --mean-stem-size, --load-capacity.	fhops.productivity.forwarder_bc.ErikssonResult

Grapple Skidder & Shovel Logger Productivity

fhops dataset estimate-productivity --machine-role grapple_skidder wires directly into fhops.productivity.grapple_bc.estimate_grapple_skidder_productivity_adv6n7() and related helpers. Use these presets to shortcut tuning:

--grapple-skidder-model	When to use	Key CLI options	API reference
adv6n7	FPInnovations ADV6N7 long-distance grapple skidding with optional decking/utilisation overrides.	--skidder-extraction-distance, --skidder-adv6n7-decking-mode, payload/utilisation/delay overrides.	fhops.productivity.ADV6N7SkidderResult
han-whole-tree / han-lop-and-scatter	Han et al. (2018) beetle-kill salvage cycles where piece size and empty/loaded distances are known.	--pieces-per-cycle, --piece-volume-m3, --skidder-empty-distance, --skidder-loaded-distance.	fhops.productivity.grapple_bc.estimate_grapple_skidder_productivity_han2018()
adv1n12-fulltree / adv1n12-two-phase	Advantage 1N12 thinning presets (one- vs two-phase). Use when you only know extraction distance.	--skidder-extraction-distance (other parameters default from the preset).	fhops.productivity.estimate_cable_skidder_productivity_adv1n12_full_tree()
shovel-logger (serpentine)	Sessions & Boston shovel-logging with directional slope/bunching multipliers.	--shovel-passes plus swing/payload inputs; defaults come from fhops.productivity.ShovelLoggerSessions2006Inputs.	fhops.cli.dataset._evaluate_shovel_logger_result() (see CLI docstrings)

Skyline & Helicopter Productivity

Running skyline presets share parameters with the cable helpers inside fhops.productivity.cable_logging. The CLI docstrings now list every parameter, but a few quick rules help:

Skyline model	When to use	Key CLI options	API reference
tr125-single / tr125-multi	Skylead C40 single-span or intermediate-support operations (payload defaults to 1.6 m³).	--slope-distance, --lateral-distance (multi-span pulls TN258 lateral warnings automatically).	fhops.productivity.estimate_cable_yarder_productivity_tr125_single_span()
tr127-block{1..6}	FPInnovations TR127 blocks with dual lateral distances and logs-per-turn requirements.	--lateral-distance + --lateral-distance-2 and (Blocks 5–6) --num-logs.	fhops.productivity.estimate_cable_yarder_productivity_tr127()
fncy12-tmy45	Thunderbird TMY45 with Mini-Mak II supports (pulls Cat D8/Timberjack ratios + TN258 envelopes).	--fncy12-variant only; payload/shift hours come from fhops.productivity.estimate_tmy45_productivity_fncy12().	fhops.productivity.estimate_tmy45_productivity_fncy12()
hi-skid / tn173-* / adv5n28-*	Short-span skyline conversions (Hi-Skid truck, TN173 micro yarders, ADV5N28 helicopter-to-skyline upgrades).	--pieces-per-cycle, --piece-volume-m3, optional --payload-m3 (defaults seeded per system ID).	fhops.productivity.cable_logging (see individual helper docstrings)

For helicopter workflows, fhops dataset estimate-productivity --machine-role helicopter_longline maps directly to fhops.productivity.estimate_helicopter_longline_productivity(). Select a preset with --helicopter-preset to seed the distances and payload weights documented in the helper docstring.

Forwarder Productivity Models

Cut-to-length systems rely on the forwarder helper stack (fhops.productivity.forwarder_bc) when fhops dataset estimate-productivity or fhops dataset estimate-forwarder-productivity is invoked. Pick the regression that matches the stand context:

• ghaffariyan-small / ghaffariyan-large – ALPACA plantation thinning data (14 t and 20 t forwarders). Requires --extraction-distance plus either --slope-class or --slope-factor. Use these when analysing Australian-style commercial thinning or when you need quick distance/slope sensitivity without payload inputs.

• kellogg-sawlog / kellogg-pulpwood / kellogg-mixed – FMG 910 regression from Kellogg & Bettinger (1994) for western Oregon CTL operations. Supply --volume-per-load, --distance-out, --travel-in-unit, and --distance-in. Ideal for moderate (≤350 m) extraction distances and when you must distinguish sawlog vs. pulp payloads.

• adv6n10-shortwood – FPInnovations ADV6N10 shortwood model for boreal multi-product sorting. Requires --payload-per-trip, --mean-log-length, --travel-speed, --trail-length, and --products-per-trail. Use this when you have explicit payload/log-length targets or need to quantify the productivity penalty of sorting several products per trail.

• adv1n12-shortwood – Advantage Vol. 1 No. 12 extraction-distance regression for the Valmet 646 forwarder (formula stored in data/productivity/forwarder_skidder_adv1n12.json). Provide --extraction-distance (m); the helper reports the decay curve (8.4438·e^(−0.004·d)) plus the usual PMH0 output. Use when road spacing dictates forwarding distances between roughly 150–600 m and you want the published optimum (~400 m) reflected in your sensitivity runs.

• eriksson-final-felling / eriksson-thinning – Eriksson & Lindroos (2014) Swedish follow-up dataset covering >700 CTL forwarders. Provide --mean-extraction-distance (m), --mean-stem-size (m³), and --load-capacity (m³). The final-felling variant reflects larger stems and long forwarding distances; the thinning variant should be used when stem sizes stay below ~0.2 m³. These regressions omit explicit slope factors, so treat steep ground as an open action item until BC-specific multipliers are published.

• laitila-vaatainen-brushwood – Ponsse Buffalo Dual harwarder regression from Laitila & Väätäinen (2020). Supply --harvested-trees-per-ha, --avg-tree-volume-dm3, and --forwarding-distance. Optional knobs --harwarder-payload (default 7.1 m³) and --grapple-load-unloading (default 0.29 m³) match the study’s payload assumptions. This helper is best suited for whole-tree/brushwood clean-up, light salvage, or plantation corridor clearing where a single machine fells, accumulates, and shuttles material to the roadside.

The commands surface identical arguments under both the dedicated forwarder CLI and the general estimate-productivity --machine-role forwarder path, so the same guidance applies to synthetic generators, scenario QA, and KPI workflows.

Harvest-system templates now include two ADV1N12 presets:

• thinning_adv1n12_forwarder – sets the forwarder model to adv1n12-shortwood with a default 350 m extraction distance (mirroring the Advantage curves) and uses the Valmet 646 machine-rate entry (owning 18.85 $/SMH + operating 45.73 $/SMH in 1999 CAD, automatically inflated to the CLI target year). Selecting this system via --harvest-system-id or dataset blocks pre-fills the forwarder CLI inputs and –show-costs reports the CPI-adjusted Appendix 1 cost split.

• thinning_adv1n12_fulltree – wires the lop-and-scatter Timberjack 240 skidder preset (see the grapple-skidder section) so planners can contrast the shortwood vs. full-tree options without retyping the Advantage extraction distances.

Roadside Processor Productivity Models

Landing processor and loader coverage cheat sheet

The following tables summarise which preset to grab for common BC case-study scenarios. Use them as a quick index before diving into the per-model notes.

Landing processor presets (region + utilisation snapshot)

Model flag	Region / primary deployment	Key inputs & utilisation	Notes
adv5n6	Coastal BC – Madill 3800 + Waratah HTH624 (loader-forwarded cold decks vs. grapple-fed hot decks).	Stem source + processing mode select PMH/SMH and CPI-scaled $/m³ (≈0.73 utilisation baked in).	Use when replaying ADV Vol. 5 No. 6 landings; hot decks mirror yarder waits, low-volume mode captures small swing camps.
tn166	Interior BC – Denis D3000 telescopic boom stroke processor (grapple-yarded vs. right-of-way vs. mixed shift).	Scenario flag picks utilisation (≈0.78–0.87) and CPI-normalised costs.	Handy for stroke-processor partial cuts or right-of-way clean-up.
tr87 / tr106 / tn103	Legacy BC roadside CTL / partial cut (TJ90 twin processors, Steyr KP40 variants, Cat DL221 stroke).	Scenario selector prints PMH, SMH, and CPI-normalised costs; no extra numeric inputs.	Pair coastal TR-87/TN-103 for Vancouver Island/Haida Gwaii studies and TR-106 for Cariboo shelterwoods.
berry2019	Kinleith, NZ (purpose-built processor baseline). Useful surrogate when BC dimensions match Berry’s 1–3 m³ stems.	Needs piece size, tree-form, crew multiplier, delay multiplier (default 0.91); carrier option auto-adjusts utilisation.	Serves as general landing processor baseline when local BC data is unavailable but piece-size inputs are known.
labelle2016/2017/2018/2019_*	Eastern Canadian / Bavarian hardwoods (DBH or recovered volume regressions, PMH₀ only).	Require DBH or recovered volume plus species/treatment/variant; always supply a utilisation multiplier for BC use.	Treat outputs as delay-free references for export markets or hardwood-heavy BC trials; not calibrated for conifer landings.
visser2015	NZ cable landings (Cat 330DL + Waratah) focused on log-sort complexity.	Requires piece size (1–3 m³) and log-sort count; utilisation multiplier defaults to 0.91.	Quick way to quantify productivity loss when chasing >5 log sorts at coastal BC cable yards.
hypro775	Tractor-mounted processor (Castro Pérez 2020; Zurita 2021) for thin-wood landings.	No extra inputs; default utilisation ≈0.73 with noise/ergonomic warnings.	Use for small private-land salvage or community forest pilots where capital limits rule out large processors.
bertone2025 / spinelli2010	European excavator processors at cable landings (Italian Alps) with heavy yarder waits.	Need DBH/height/log count/piece size (Bertone) or tree volume + machine power + slope + stand info (Spinelli); utilisations baked in.	Helpful for steep-slope BC pilots with winch-assist yarders or when benchmarking against European CTL standards.
borz2023 / nakagawa2010	Landing harvester-as-processor (Romania) and excavator-based processor (Hokkaido thinning).	Borz needs no inputs (optionally piece size); Nakagawa accepts DBH or per-tree volume plus a delay multiplier.	Use to bound harvester bucking scenarios or shovel-processor deployments when only tree dimensions are known.

Loader / hoe-forwarder presets (BC defaults)

Model flag	Region / primary deployment	Key inputs & utilisation	Notes
tn261	Interior BC loader-forwarder (Hoe-chucking) reference.	Requires piece size, forwarding distance, slope %, bunched flag, and delay multiplier (utilisation default = delay-free).	Baseline for hoe-chucking between road and landing; pair with TN285 trail spacing notes.
adv5n1	Coastal BC Madill 3800 shovel-forwarder (ADV5N1 Figure 9).	Inputs: forwarding distance, slope class (0–10 % vs. 11–30 %), payload, optional utilisation override (default ≈0.93).	Captures shovel-fed loader operations for roadside decking or short forwarding legs.
adv2n26	Clambunk/hoe-forwarding combo (Trans-Gesco TG88 + JD 892D-LC) from ADV2N26.	Needs travel empty distance, stems per cycle, stem volume, utilisation (default 0.87) plus optional in-cycle delays.	CLI prints soil-disturbance stats (20.4 % trail occupancy) so planners can track disturbance budgets.
barko450	TN-46 Barko 450 heel-boom loader (ground-skid vs. cable block).	Scenario flag selects utilisation (0.79 default) and CPI-inflated machine rate; optional –loader-utilisation rescales cost per m³.	Use for live-heel truck loading at coastal BC landings; cost role flows through inspect-machine.
kizha2020	Biomass/bucket loader reference (hot vs. cold loading) from Maine study.	Mode flag sets utilisation (55 % hot vs. 7 % cold) and costs; no other inputs.	Handy for demonstrating hot-deck decoupling costs or biomass operations; not BC-calibrated but informative.

fhops.dataset estimate-productivity --machine-role roadside_processor now supports two regressions:

• --processor-model berry2019 (default) – Berry (2019) Kinleith NZ processor study keyed to piece size, tree form, and crew/utilisation multipliers. fhops dataset berry-log-grades (or --processor-show-grade-stats) dumps the digitised Appendix 13 emmeans table so you can see the per-grade mean ±2σ cycle times we extracted from the thesis image. Treat those values as qualitative cues only—they came from a screenshot, not a raw table.

  • --processor-piece-size-m3 – average piece size per stem (m³). Delay-free productivity is computed via 34.7 × piece_size + 11.3.

  • --processor-tree-form – 0 (good), 1 (poor), 2 (bad). Tree-form penalties follow Berry’s observed processing-time uplift (category 1 = +56 % time ⇒ productivity ×0.64, category 2 = +84 % ⇒ ×0.54).

  • --processor-crew-multiplier – optional operator adjustment (crew A ≈ +16 % ⇒ 1.16, crew C ≈ −25 % ⇒ 0.75, etc.).

  • --processor-delay-multiplier – utilisation factor (default 0.91 reflects delays <10 min logged in the study). Adjust it if your PMH/SMH ratio differs.

  • --processor-carrier – purpose_built (default) or excavator. Excavator carriers apply the utilisation/productivity penalty observed by Magagnotti et al. (2017) and surface the higher fuel consumption + yarder-wait behaviour noted by Nakagawa (2010) / Bertone & Manzone (2025).

  • --processor-skid-area-m2 – optional landing skid-area flag. When provided, the CLI evaluates Berry’s Figure 11 regression (≈2,600–3,700 m² range) to predict delay seconds per stem and, if you have not supplied --processor-delay-multiplier, auto-scales utilisation to reflect the larger/smaller landing. The command also prints the predicted m³/PMH hint and warns when your skid area lies outside the study range.

• --processor-model labelle2016 – Labelle et al. (2016) sugar maple study (New Brunswick) grouped by tree form quality (acceptable vs. unacceptable). Outputs are PMH₀. * --processor-dbh-cm – diameter at breast height (cm). * --processor-labelle2016-form – acceptable or unacceptable form class (matches the NHRI tree-form groupings). * --processor-delay-multiplier – optional utilisation scaling.

• --processor-model labelle2017 – Labelle et al. (2017) excavator-based CTL processor regressions (New Brunswick hardwoods). Includes two cubic polynomials and two power-law fits. * --processor-dbh-cm – diameter at breast height (cm). * --processor-labelle2017-variant – poly1, poly2, power1, or power2 (mirrors Appendix 8 table). Use the polynomial variants for large sample sizes (338/365 trees) or the power-law variants when matching the smaller subsets (42/55 trees).

• --processor-model labelle2018 – Labelle et al. (2018) Bavarian beech/spruce study (Ponsse Bear + H8) with separate regressions for rubber-tired (rw) vs. tracked (ct) processors. * --processor-dbh-cm – diameter at breast height (cm). * --processor-labelle2018-variant – rw_poly1, rw_poly2, ct_poly1, or ct_poly2 (Appendix 8 Table). Outputs are PMH₀; apply utilisation as needed.

• --processor-model labelle2019_dbh – Labelle et al. (2019) Bavarian hardwood case study (TimberPro 620-E + LogMax 7000C) using DBH polynomials per species/treatment. These regressions output delay-free PMH₀ productivity for large-diameter, hardwood-dominated stands (rare in BC but useful when deploying FHOPS abroad).

  • --processor-dbh-cm – diameter at breast height (cm).

  • --processor-species – spruce or beech (matching the reference plots).

  • --processor-treatment – clear_cut or selective_cut (group selection). Use --processor-delay-multiplier if you need to impose a local utilisation ratio.

• --processor-model labelle2019_volume – same Bavarian dataset, but keyed to recovered tree volume (m³/stem). Accepts the same species/treatment flags plus:

  • --processor-volume-m3 – recovered volume per stem (m³). The helper applies the a + b·V₆ − c·V₆² polynomials published in Appendix 8/9 and still reports PMH₀ outputs.

  • Continue using --processor-delay-multiplier for utilisation assumptions; CLI output reminds users that this model is intended for hardwood-dominated export scenarios rather than BC norms.

• --processor-model adv5n6 – FPInnovations Advantage Vol. 5 No. 6 coastal BC processor study (Madill 3800 with Waratah HTH624). Use --processor-stem-source (loader_forwarded or grapple_yarded) plus --processor-processing-mode (cold, hot, or low_volume) to match the published cold-deck loader-forwarded scenario vs. the grapple-yarded hot/cold decks. Outputs provide both PMH and SMH productivity along with the $/m³ cost figure from the report, so analysts can keep BC roadside presets tied to local landing data. Cost figures shown in the CLI are escalated to 2024 CAD using Statistics Canada CPI (Table 18-10-0005-01) so they’re comparable with newer references.

• --processor-model adv7n3 – FPInnovations Advantage Vol. 7 No. 3 (Canfor Mackenzie summer skyline settings). Set --processor-adv7n3-machine to hyundai_210 or john_deere_892 to pull the observed shift/detailed productivity plus the processor-only vs. loader-only vs. combined $/m³ costs (2004 CAD and CPI-adjusted 2024 CAD). CLI output also prints the loader task breakdown, non-processing time penalties when the processor works without a loader, and the average travel distances so you can decide when a loader is mandatory on short-wood cable decks.

  Harvest-system presets cable_running, cable_running_adv5n28_clearcut, and cable_running_adv5n28_shelterwood now default to this model (Hyundai 210LC for the base/shelterwood runs, John Deere 892 for the clearcut conversion). Invoke --harvest-system-id <id> (or point to a dataset block using those systems) and the CLI auto-selects the right ADV7N3 machine, prints the Advantage citation, and emits the CPI-adjusted processor + loader costs without retyping any flags.

• --processor-model hypro775 – HYPRO 775 tractor-mounted double-grip processor preset sourced from Castro Pérez (2020) and Zurita Vintimilla (2021). No additional predictors are needed; the helper reports mean cycle time (~45 s), gross/net trees per hour, delay-free m³/PMH (~21.4), fuel use (≈21 L/h or 0.78 L/m³), and ergonomic notes (≈85 dB(A), heavy-work cardiovascular load). Use this when modelling tractor-mounted landing processors in small-diameter thinnings or low-investment deployments.

• --processor-model bertone2025 – Bertone & Manzone (2025) excavator-processor case study serving a cable yarder landing in the Italian Alps. Provide --processor-dbh-cm, --processor-tree-height-m, --processor-logs-per-tree (2–5), and --processor-piece-size-m3 for the tree volume; the helper evaluates the published cycle-time regression and reports delay-free vs. observed productivity (≈25.9 vs. 14.8 m³/h), fuel (~16.5 L/SMH), and €-denominated owning costs with the 48 % yarder-wait penalty.

• --processor-model spinelli2010 – Spinelli, Hartsough & Magagnotti’s Italian CTL productivity standards (Forest Products Journal 60(3):226–235). Choose between harvest (in-stand) and landing-processing modes via --processor-spinelli-operation and supply the tree volume (--processor-piece-size-m3), carrier power (--processor-machine-power-kw), slope (--processor-slope-percent), stand type, carrier/head/species flags, and (for harvest) the removals/residuals per hectare. The helper applies the published accessory (14.7/29.6 %) and delay (50/21/44 %) ratios automatically and prints the per-element cycle times so you can see how slope, carrier choice, and species mix influence delay-free PMH outputs.

• --processor-model borz2023 – Borz et al. (2023) single-grip harvester bucking at a Romanian landing. No extra inputs are needed beyond the optional --processor-piece-size-m3 for reference; the helper reports the observed averages (≈21.4 m³/PMH, 18.8 m³/SMH, 0.047 PMH/m³, 0.78 L/m³ fuel, 10–11 EUR/m³ cost, 95 % recovery) so planners can benchmark “harvester-as-processor” landing operations against manual or excavator-based options.

• --processor-model nakagawa2010 – Nakagawa et al. (2010) excavator-based processor (Timberjack 746B carrier with Komatsu PC138US) working at a Hokkaido landing. Supply at least one predictor: * --processor-dbh-cm – applies the published delay-free regression 0.363·DBH^1.116 (DBH in cm). * --processor-piece-size-m3 – applies the alternative regression 20.46·V^0.482 (tree volume in m³). * --processor-delay-multiplier – optional utilisation factor so you can fold landing waits into the PMH₀ baseline (defaults to the global processor multiplier; set to 1.0 to stay delay-free). CLI output states which regression ran, prints the delay-free m³/PMH alongside your utilisation-adjusted result, and cites the Japanese thinning context so you can benchmark excavator-based landing processors handling ≈0.25 m³ stems.

Note

All cost snippets in this reference (and in the CLI) call fhops.costing.machine_rates.compose_default_rental_rate_for_role() to pull owning/operating/repair components from the bundled OpCost tables, then inflate the totals to fhops.costing.inflation.TARGET_YEAR CAD via fhops.costing.inflation.inflate_value(). Telemetry summaries reuse fhops.telemetry.machine_costs.build_machine_cost_snapshots() so KPI exports and reports print the same numbers shown in the CLI --show-costs panels.

Optional Dataset Cookbook

Use these quick recipes to exercise optional dataset columns and CLI helpers that accompany the bundled scenarios.

Med42 / Large84 cost overlays

• Copy examples/med42 or examples/large84 into a fresh workspace and add harvest_system_id to data/blocks.csv whenever you want deterministic system assignments. The IDs listed earlier in this guide align with the CLI help surfaced by fhops.cli.dataset.inspect_machine(), so you can paste the table output directly into a thesis appendix.

• Run fhops dataset inspect-machine examples/med42/scenario.yaml --machine loader_1 --show-costs to print the CPI-adjusted owning/operating/repair breakdown for any machine in the scenario. This command uses fhops.telemetry.machine_costs.build_machine_cost_snapshots(), so solver telemetry and CLI output stay in sync.

• Need a $/m³ estimate with local utilisation? fhops dataset estimate-cost --dataset examples/med42 --machine loader_1 --utilisation 0.85 invokes fhops.cli.dataset.estimate_cost_cmd() to blend the rental-rate components with utilisation (and optional Lahrsen inputs). Add --road-machine cat_d8h --road-length-m 250 to include a TR-28 subgrade surcharge (see the road-construction recipe below).

Partial-cut / mixed system assignments

• Partial-cut profiles live in fhops.productivity.partial_cut_profiles and can be referenced by adding a partial_cut_profile column to data/blocks.csv. The CLI echoes the chosen profile whenever you pass --partial-cut-profile to fhops.cli.dataset.estimate_productivity_cmd(), making it easy to cite SR-109/SR-54 penalties in a report.

• To reproduce the Block 5 patch-cut scenario from TR-127, call fhops dataset estimate-productivity --machine-role grapple_yarder --model tr127-block5 --partial-cut-profile sr109_green. The helper prints the calibration warnings, dual lateral distances, and CPI-adjusted cost references described earlier.

• Micro-yarders from TN173 follow the same pattern: set harvest_system_id=cable_micro_ecologger (or the other micro presets) in your dataset and the CLI auto-selects the matching preset when running fhops dataset estimate-productivity.

Road construction add-ons

• fhops dataset tr28-subgrade lists the Caterpillar/Poclain/American machines digitised from TR-28. Use fhops dataset tr28-subgrade --detail to view cycle times, unit costs, and soil profile IDs before wiring them into a scenario.

• fhops dataset estimate-cost supports TR-28 surcharges through --road-machine / --road-length-m and the optional --road-soil-profile flags. The command calls fhops.cli.dataset.estimate_cost_cmd(), which dispatches to fhops.cli.dataset.estimate_tr28_road_cost() and prints the ADV4N7 compaction warnings so the full cost per cubic metre stays transparent.

• Scenarios can embed a road_construction table (see tests/fixtures/regression/road_construction.yaml). When you pair --dataset/--machine with fhops dataset estimate-cost, the CLI automatically loads the matching road entry, applies the correct machine slug, and surfaces the CPI-adjusted totals—ideal for med42/large84 experiments where you want to toggle spur-road budgets without editing the CLI command each time.

• --processor-model visser2015 – Visser & Tolan (2015) NZ cable-yarder landings (Cat 330DL + Waratah HTH626) comparing 5/9/12/15 log sorts. The helper interpolates the published piece-size curves (1–3 m³ stems) and reports both the delay-free m³/PMH and the delta versus the 5-sort baseline.

  • --processor-piece-size-m3 – mean stem volume per piece (must fall within the 1–3 m³ study range).

  • --processor-log-sorts – number of log sorts to cut (5, 9, 12, 15). CLI output includes the gross-value and value-per-PMH figures from the paper (2.0 m³ reference, 2014 USD).

  • --processor-delay-multiplier – optional utilisation scaling (default 0.91). The CLI applies this multiplier to the Visser delay-free values so you can simulate longer shifts or landing congestion.

• --processor-model tn103 – FERIC TN-103 Caterpillar DL221 evaluation in coastal old growth. Pick --processor-tn103-scenario (area_a_feller_bunched, area_b_handfelled, combined_observed, or combined_high_util) to reflect how well windrows are prepared and whether the 73 % utilisation improvement has been implemented. The helper reports trees/PMH, m³/PMH, and the published $/m³ and $/tree costs; CLI output also notes windrow prep guidance so analysts can match conditions to their landing prep.

• --processor-model tr87 – FERIC TR-87 roadside CTL case study (two Timberjack TJ90 processors plus LL229 loader). Use --processor-tr87-scenario to choose day shift, night shift, the observed combined average, or the “wait-for-wood removed” system scenario (240 m³/shift target). Outputs list trees per PMH, m³/PMH/SMH, utilisation, and the $2.67/m³ processing cost so you can benchmark historic TJ90 fleets or sanity-check inland roadside plans.

• --processor-model tr106 – FERIC TR-106 lodgepole pine shelterwood study near Williams Lake. Pick --processor-tr106-scenario (Case 1187 Oct–Nov vs. Feb, or the Steyr KP40 carriers on Cat 225 / Link-Belt L-2800 / Cat EL180) to see the published PMH vs. net-PMH productivity, stems per hour, cycle minutes, utilisation, and CPI-scaled cost metadata. Handy when you need an interior partial-cut preset to complement the coastal TR-87 helper.

• --processor-model tn166 – FERIC TN-166 stroke-processor study (Denis D3000 telescopic boom) from interior BC. Choose --processor-tn166-scenario (grapple_yarded, right_of_way, or mixed_shift) to flip between the detailed timing splits or the overall shift-level average. The helper reports PMH/SMH productivity, stems per hour, and the published $/m³ + $/stem values. Pick this preset when you need an interior stroke-processor baseline rather than the hardwood-focused Labelle regressions. Costs are also escalated to 2024 CAD via the same Statistics Canada CPI series.

• fhops dataset unbc-hoe-chucking – reference command that prints the UNBC (Renzie 2006) Table 33 hoe-chucking shift summary (time, volume, productivity, observed/weighted $/m³ for group selection, group retention, and clearcut treatments). Useful when budgeting manual hoe-chucking support around landing processors in cedar–hemlock partial cuts.

• fhops dataset adv2n21-summary –treatment partial_cut_2 – surfaces the ADV2N21 Timberjack 1270/1010 partial-cut costs (1997 CAD $/m³), CPPA terrain classes, and pre/post stand metrics for each patch/partial/ clearcut block. Handy when you need to cite the +11 %…+78 % cost penalties for patch cuts or quantify the wildlife-corridor snow constraints before tuning forwarder defaults.

• --processor-automatic-bucking – Optional switch for Berry/Labelle helpers that applies the +12.4 % delay-free productivity multiplier from Labelle & Huß (2018, Silva Fennica 52(3):9947) and prints the associated €3.3/m³ revenue delta (2018 EUR) for traceability. The flag is ignored for the table-driven FPInnovations presets because those studies didn’t use the on-board bucking optimizer.

CLI output reports the base delay-free productivity, the applied multipliers, and the utilisation-adjusted m³/PMH so costing workflows can decide which value to pass downstream. Remember that the Labelle models were calibrated outside BC—treat them as hardwood export presets and rely on the Berry helper for mixed or conifer-dominated BC blocks.

Burned-Timber Salvage Guidance (ADV1N5)

fhops.dataset adv1n5 does not exist because the study is qualitative, but the skyline/forwarder presets inherit the following cautions from ADV1N5 (Alberta 2000 salvaging workshop + case studies). Use these checklists whenever you point cable_running*, ground_fb_*, or custom salvage systems at burned timber:

• Log preparation – buck out catfaces, scars, rotten butts, and heavily charred sections; raise the acceptable minimum top diameter; sort log decks by burn severity so dirty sorts can be chipped or wasted separately.

• Debarking – double-ring (counter-rotating) debarkers or a Cambio + Nicholson tandem remove charcoal reliably; alternate cutter/winter tips, slow the infeed ≥10 %, and recycle logs for multiple passes when residue remains.

• Charcoal dust – spray logs pre-debarker when temperatures allow, otherwise install vac/baffle systems, pressurize MCC rooms, and mandate PPE + aggressive greasing schedules to keep dust from destroying rings, bearings, and control panels.

• Priorities – harvest small-diameter or severely burned stands first (moisture loss + insect attack), and route environmentally sensitive slopes/gullies to grapple yarders (e.g., the Madill 122 case delivered 300–400 m³/10 h once stems were bunched parallel to the slope).

• Processing capacity – deploy portable mills at satellite yards or in-woods chip plants for small stems when permanent mills cannot expand fast enough; burn or isolate slab/char waste so pulp furnish stays clean. Use --salvage-processing-mode (standard_mill | portable_mill | in_woods_chipping) when calling ground_salvage_grapple or cable_salvage_grapple to print the appropriate reminder in the CLI output, and store the selected mode in blocks.csv via the salvage_processing_mode column so downstream scenario tooling and telemetry inherit the same warning set automatically.

Document any of these mitigations in scenario metadata (telemetry note, –harvest-system-id description, etc.) when you model salvage corridors so downstream QA knows why costs/utilisation differ from green-timber runs. The new ground_salvage_grapple and cable_salvage_grapple harvest systems simply bundle the same checklist so you can select them via --harvest-system-id without rewriting the warnings.

Loader-Forwarder Productivity Models

fhops.dataset estimate-productivity --machine-role loader now supports three literature-backed presets:

• --loader-model tn261 (default) – the Vancouver Island loader-forwarder study (FERIC TN-261, 1994). Provide:

  • --loader-piece-size-m3 – mean stem volume per turn (m³).

  • --loader-distance-m – external distance from deck to the farthest stem (m).

  • --loader-slope-percent – approximate slope (%) along the forwarding direction. Positive values represent adverse (uphill) travel; negative values represent favourable (downhill) travel.

  • --loader-bunched/--loader-hand-felled – indicates whether stems are mechanically bunched/aligned or hand-felled/scattered (hand-felled defaults to a 0.90 multiplier).

  • --loader-delay-multiplier – optional utilisation factor (default 1.0 because TN-261 detailed timing was delay-free).

  The helper uses a log-linear fit (piece-size exponent ≈0.41, distance exponent ≈−0.60) with gentle slope adjustments (≈3 % penalty per +10 % uphill, ≈1.5 % bump per −10 % downhill, clamped between 0.6 and 1.15). CLI output reports the delay-free vs. utilisation-adjusted productivity along with the applied multipliers so scenario costing flows can pick the appropriate value.

• --loader-model adv2n26 – the Kosicki (2001) clambunk/hoe-forwarding trial (FPInnovations ADV-2 No. 26) with a Trans-Gesco TG88, John Deere 892D-LC loader-forwarder, and Link-Belt landing loader. You can use the study defaults or override them with:

  • --loader-travel-empty-m – travel empty distance (m) used in Equation 1 (default 236 m).

  • --loader-stems-per-cycle – stems per cycle (default 19.7).

  • --loader-stem-volume-m3 – average stem volume per piece (default 1.52 m³/stem; payload is derived from stems × volume unless you supply a custom payload via the helper).

  • --loader-utilisation – PMH/SMH ratio (default 0.77 from the shift-level study).

  • --loader-in-cycle-delay-minutes – optional override for in-cycle delay minutes (<10 min delays); omit it to use the published 5 % ratio from Figure 9.

  CLI output shows the delay-free vs. total cycle minutes, payload per turn, and both m³/PMH and m³/SMH so you can pair the loader helper with the TG88 clambunk costs. This model is especially handy for “hot logging” corridors where a loader-forwarder supports a clambunk or skyline operation. The ADV2N26 study also documented soil disturbance: 20.4% of the harvested area was covered by unbladed trails (0.54 ha first-class at 6.2 m mean width with 91% exposed mineral soil, 0.87 ha second-class at 4.7 m / 39%, 0.49 ha third-class at 4.0 m / 14%). FHOPS prints these stats as a reminder when you select adv2n26 so you can mirror the disturbance footprint in planning or compliance reports.

• --loader-model adv5n1 – the Madill 3800 loader-forwarder regression from ADV-5 No. 1 (Figure 9). The slope-class regressions (0–10 %, 11–30 %) were manually digitised from the report so you can recover the original linear fits:

  • --loader-distance-m – forwarding distance (m) for the regression.

  • --loader-slope-class – 0_10 (baseline) or 11_30 (adds the 18 % penalty discussed in the text).

  • --loader-payload-m3 – payload per cycle (default 2.77 m³/cycle).

  • --loader-utilisation – utilisation multiplier (default 0.93 per the ADV5N1 shift-level study).

  CLI output reports the linear coefficients (intercept/slope), cycle time (minutes), delay-free m³/PMH, and utilisation-adjusted m³/SMH so corridor costing flows can adopt the ADV5N1 defaults immediately.

• --loader-model barko450 – FERIC TN-46 Barko 450 live-heel loader preset for interior roadside decks. No predictors are required; instead pick --loader-barko-scenario (ground_skid_block for the grapple-skid trial or cable_yard_block for the Washington 78 yarder decks) to show the published ≈658 m³/shift averages, availability (96 %), utilisation (79 %), and the ~17 % truck-wait/move penalty. Supplying --loader-utilisation rescales the Barko production and $/m³ outputs so you can model different truck-supply assumptions. This preset now draws its $257.26/shift (1986 CAD) cost baseline directly from the TR‑73/TN‑64/TN‑51 roadside sort studies; run with --show-costs to see the CPI-inflated owning/operating/repair breakdown that feeds the costing CLI.

• --loader-model kizha2020 – Loader utilisation/delay reference from Kizha et al. (2020) northern California biomass landings (Thunderbird 840W supporting a Madill yarder). Use --loader-hot-cold-mode (hot = integrated sawlog/biomass loading, cold = decoupled biomass-only loading) to see how utilisation collapses when the roll-off truck fleet is undersized. CLI output reports the 55 % vs. 7 % utilisation, the dominant delay contributors, and the $/PMH penalty attributable to waiting so you can size truck fleets/landing space before committing to a hot or cold biomass workflow.

Loader harvest-system overrides

Referencing a harvest-system template (--harvest-system-id or --dataset/--block-id) now auto-fills loader inputs the same way shovel, skyline, and helicopter presets work. The registry currently includes:

• ground_fb_skid – feller-buncher → grapple-skidder sequences now pin the ADV6N7 Caterpillar 535B preset (85 m extraction distance, loader-supported decking mode, 0.4 support ratio, 7.69 m³ payload, 0.12 min delays, 0.85 utilisation) and the loader job still defaults to the TN-261 helper with 1.05 m³ logs, 115 m forwarding distance, 8 % adverse slope, bunching enabled, and a 0.95 delay multiplier.

• ground_fb_loader_liveheel – same chain as ground_fb_skid but the loader job switches to the Barko 450 preset with --loader-barko-scenario ground_skid_block so coastal/interior live-heel conversions inherit the TN-46 utilisation warning (96 % availability, 79 % utilisation, 17 % truck waits) automatically while the grapple-skidder leg still leverages the ADV6N7 defaults.

• ground_salvage_grapple – pairs the ADV6N7 grapple-skidder preset with the ADV1N5 salvage guidance (buck out catfaces, raise minimum tops, double-ring debarkers, charcoal-dust controls). Use this when modelling burned-timber corridors that follow the salvage checklist in the section below.

• ground_fb_shovel – coastal shovel-logging corridors pick the ADV5N1 helper (0–10 % slope class) with a 90 m forwarding distance plus the published payload/utilisation defaults (2.77 m³/cycle, 0.93 PMH/SMH).

• steep_tethered – tethered hot-logging layouts wire in the ADV2N26 clambunk regression with 320 m travel-empty distance, 18 stems per cycle, 1.35 m³ stem volume, and a 0.77 utilisation factor.

Overrides use the same keys as the CLI flags (loader_model, loader_piece_size_m3, loader_distance_m, loader_slope_percent, loader_bunched, loader_travel_empty_m, loader_slope_class, etc.). When one of these presets applies, fhops.dataset prints a dimmed notice and telemetry captures the originating harvest system so downstream costing/solver workflows retain provenance.

Coordinating with forwarder / skidder models

Loader helpers should inherit the same assumptions as the upstream primary-transport phase so cost rollups stay consistent:

• Ground-based ghost trails (TN-261 + TN-285). The TN-261 loader-forwarder study was paired with ghost-trail spacing and pre-bunched decks described in FPInnovations TN-285. If you call the TN-261 helper outside the harvest-system presets, mirror the same trail pattern you fed to the grapple-skidder helper (TrailSpacingPattern.SINGLE_GHOST_18M corresponds to TN-285’s 18 m ghost trail layout). That keeps loader forwarding distance aligned with the average skid-trail spacing and prevents double counting of trail penalties.

• Clambunk + hoe-forwarding chains (ADV2N26 + PNW-RP-430). The ADV2N26 helper represents the Trans-Gesco TG88 / JD 892D-LC tandem documented again in Lambert & Howard’s PNW Research Paper 430. When modelling “hot logging” corridors where the clambunk hands off to a loader-forwarder, use the same stems-per-cycle and payload splits for both the clambunk helper and ADV2N26 loader helper so the landing payload is conserved. PNW-RP-430 also lists travel-speed penalties for steep trail networks; use those multipliers to adjust the ADV2N26 travel-empty distance before computing loader productivity.

• Shovel-fed landings (ADV5N1). ADV5N1 assumes the landing loader is fed by shovel logging or short skyline corridors. If the prior phase used the shovel logger helper, carry the same payload-per-swing and utilisation targets into the loader defaults so only one phase absorbs the congestion penalty.

These references are called out in notes/reference_log.md (entries for TN285.PDF and pnw_rp430.pdf) so analysts can trace the provenance when tuning harvest systems.

Grapple Skidder Productivity Models

Ground-based full-tree systems can now call the Han et al. (2018) regressions via fhops.dataset estimate-productivity --machine-role grapple_skidder:

• han2018-lop-scatter – delay-free cycle time for Tigercat 615C skidding delimbed logs in lop-and-scatter salvage. Requires --skidder-pieces-per-cycle (logs/turn), --skidder-piece-volume (m³/log), --skidder-empty-distance (m), and --skidder-loaded-distance (m). Payload is derived automatically and converted into m³/PMH0.

• han2018-whole-tree – same machine/stand, but moving whole trees to roadside decks. Replace the log count with trees per turn (same CLI flag) and supply average tree volume.

Both helpers report cycle time, payload per turn, and predicted productivity so analysts can adjust the distance or payload assumptions directly.

• adv1n12-fulltree / adv1n12-two-phase – Delay-free extraction curves from Advantage Vol. 1 No. 12 (see data/productivity/forwarder_skidder_adv1n12.json). Supply --skidder-extraction-distance (m) and the CLI evaluates the exponential (integrated lop-and- scatter crew) or logarithmic (decoupled second-phase skidder) regression. Use these when spacing roads 100–350 m apart in commercial thinnings and you want the published productivity-distance relationships without retyping the equations.

• adv6n7 – Caterpillar 535B grapple skidder regression (Advantage Vol. 6 No. 7). Provide --skidder-extraction-distance plus optional --skidder-adv6n7-* flags (decking mode, payload, utilisation, delay, support ratio). Defaults mirror the Englewood case study (7.69 m³ turns, 0.85 utilisation, 0.12 min delay, loader support ratio 0.4). The CLI prints CPI-adjusted skidding cost and combined skid+deck cost so you can compare directly with loader-forwarding baselines.

• adv1n35 – Owren 400 hydrostatic yarder regression (Advantage Vol. 1 No. 35). Provide --grapple-yard-distance-m plus the new knobs --grapple-lateral-distance-m and --grapple-stems-per-cycle (defaults 11 m and ≈2.6–2.8 stems/turn). Optional --grapple-in-cycle-delay-minutes lets you override the observed 0.69 min delay allowance. The CLI reports the assumed values and productivity so you can mirror the 200–350 m two-span corridors featured in the study when exploring interior skyline layouts.

• adv1n40 – Madill 071 running/scab skyline regression (Advantage Vol. 1 No. 40). Supply --grapple-yard-distance-m (default 103 m) and optional --grapple-in-cycle-delay-minutes (default 1.10 min). Turn volume defaults to 2.9 m³ unless overridden. Designed for downhill group-selection blocks in ICHvk1; the helper prints the CPI-adjusted $/m³ yarding cost from the publication.

Trail spacing and decking multipliers (from FPInnovations TN285 and ADV4N21) can be applied via --skidder-trail-pattern (narrow_13_15m | single_ghost_18m | double_ghost_27m) and --skidder-decking-condition (constrained_decking | prepared_decking). The defaults follow TN285’s finding that 13–15 m trail spacing drove 20–40% higher extraction costs (≈25% productivity penalty) while double-ghost 27–30 m layouts serve as the baseline. ADV4N21 reported decking-time reductions from 1.33 to 0.60 min/cycle after clearing the landing; we translate that ~16% cycle-time penalty into the constrained_decking multiplier. Analysts can stack these with an optional --skidder-productivity-multiplier when site-specific data is available.

Travel-speed assumptions now support GNSS data via --skidder-speed-profile:

• legacy (default) – use the Han et al. (2018) regression coefficients for empty/loaded travel.

• gnss_skidder – replace the distance terms with the GNSS median speeds observed for cable skidders in Zurita & Borz (2025) (≈4.1 km/h empty, 3.9 km/h loaded).

• gnss_farm_tractor – same GNSS dataset but using the farm-tractor skidders (≈5.1 km/h empty, 3.6 km/h loaded).

CLI output notes which profile was used, and harvest-system templates can override the profile the same way they do for trail/decking multipliers.

The new ADV1N12 options ignore the trail/decking multipliers (the regressions already embed those effects). When the CLI detects an ADV1N12 model it switches the output table to a condensed format that simply lists the model, extraction distance, and predicted m³/PMH for the selected regression.

Select --harvest-system-id thinning_adv1n12_fulltree to auto-populate the adv1n12-fulltree model, a representative 225 m extraction distance, and the Timberjack 240 machine-rate entry (12.33 $/SMH owning + 63.28 $/SMH operating in 1999 CAD, CPI-adjusted inside --show-costs).

Harvest-system templates can now supply these overrides automatically: pass --harvest-system-id <system_id> (default registry or your scenario’s definitions) or --dataset <scenario-path> --block-id <block> to pull the block’s harvest system and apply its productivity_overrides for the grapple skidder job. The command prints when such defaults are applied so you can verify which template influenced the result.

Grapple Yarder Productivity Models

Cable-running systems can now call the grapple yarder regressions bundled in fhops.dataset estimate-productivity --machine-role grapple_yarder:

• sr54 – MacDonald (1988) SR-54 regression for a Washington 118A grapple yarder on mechanically bunched wood. Requires --grapple-turn-volume-m3 and --grapple-yard-distance-m and includes the minor delay allowance from Table 10.

• tr75-bunched – Peterson (1987) TR-75 Madill 084 regression for mechanically bunched second growth. Uses the same CLI inputs; the helper applies the published outhaul/inhaul coefficients plus a fixed hook/unhook allowance.

• tr75-handfelled – Madill 084 handling hand-felled timber (TR-75 hand-felled regression). Use this when you need to reflect lower payload control or hand-bunched turns.

• tn157 – Observed productivity/costs from TN-157 (Cypress 7280B swing yarder + Hitachi UH14 backspar). Select a case with --tn157-case (combined default, or 1–7 to pull an individual study). The helper reports the observed turn volume, yarding distance, m³/PMH, and both 1991 CAD and CPI-inflated $/m³ so costing workflows can drop the Cypress preset in immediately. Adding --show-costs now references the grapple_yarder_cypress7280 machine-rate entry (FERIC TN-157 Appendix II) so the CLI prints the BC swing-yarder owning/operating split inline; inspect-machine --machine-role grapple_yarder_cypress7280 surfaces the same table for planning docs.

• tn147 – Observed productivity/costs from TN-147 (seven Madill 009 highlead case studies around Lake Cowichan). Pick a case with --tn147-case (combined default) to pull the corresponding logs/turn, turn volume, and 1989 CAD cost per log/m³; the CLI inflates the costs to 2024 CAD alongside the original values and, when you add --show-costs, it now references the TN-147 machine-rate entry grapple_yarder_madill009 so the owning/operating split matches the Madill 009 helper automatically.

• tr122-extended / tr122-shelterwood / tr122-clearcut – Observed Washington SLH 78 running-skyline productivity from TR-122 (Roberts Creek alternative silviculture project). These presets don’t require new inputs: the CLI reports the published cycle volume, m³/PMH, and the yarder/loader/labour cost breakdowns (1996 CAD plus CPI-inflated 2024 CAD equivalents) so you can reference the extended-rotation vs. shelterwood vs. clearcut costs.

• adv5n28-clearcut / adv5n28-shelterwood – Long-distance skyline conversions from ADV5N28 (Madill 071 + Acme 200 Pow’-R Block carriage substituting helicopter plans near Lillooet). These presets pull the observed turn volume, 375–725 m downhill yarding distances, m³/PMH, and the 2002 CAD costs for both the actual study and the projected skyline-vs-helicopter scenarios so you can quantify the savings relative to the $60/m³ heli baseline. When you add --show-costs (or call inspect-machine --machine-role grapple_yarder_adv5n28) the CLI now references the ADV5N28 Madill 071 owning/operating split (289.77 $/SMH total from the Appendix II cost table) instead of the generic swing-yarder placeholder, keeping the skyline costing consistent with the study.

Every helper prints the assumed turn volume, yarding distance (when available), and resulting m³/PMH. Dataset-driven presets (TN-147/TN-157/TR-122) also surface the observed case name and CPI-adjusted costs so you can drop them directly into costing workflows without reopening the PDFs. The cable_running harvest system autopopulates these inputs when you pass --harvest-system-id cable_running (or reference a dataset block using that system), selecting the tr75-bunched model with representative payload/distance values and printing a confirmation when the defaults are applied. Use cable_running_adv5n28_clearcut or cable_running_adv5n28_shelterwood when modelling the ADV5N28 long-span conversions—those presets swap in the new helper automatically and reuse the same skyline defaults, so you can flip helicopter blocks to skyline without retyping payload/cost assumptions.

Short-span presets cable_micro_ecologger, cable_micro_gabriel, cable_micro_christie, cable_micro_teletransporteur, and cable_micro_timbermaster now drive the TN173 skyline models (crew size, pieces/turn, and the correct machine-rate role) automatically, so selecting one of those systems drops the matching Eastern Canada helper into the CLI without retyping parameters. The cable_micro_hi_skid preset does the same for the FNG73 Hi-Skid truck—selecting that harvest system funnels --model hi-skid defaults into the skyline CLI, and you can append --hi-skid-include-haul when you want the reported productivity to fold in the 30 min haul/unload leg per 12 m³ load.

Use cable_salvage_grapple when those skyline corridors are salvaging burned timber: it mirrors the cable_running payload defaults (TN-157 combined preset + ADV7N3 deck costs) but tags the scenario as a salvage operation so the ADV1N5 cautions (parallel bunching, grapple yarding rough ground, charcoal controls) are baked into the harvest-system notes and telemetry.

Grapple harvest-system presets

Grapple harvest-system summary

Harvest system ID(s)	Linked model(s)	Default inputs	Cost role / notes
cable_running / cable_running_tn157_combined	tn157 + mcneel-running skyline swap	3.8 m³ turns, 83 m yarding distance, ≈3 pcs/turn, Douglas-fir 52.5 cm manual falling	grapple_yarder_cypress7280 when --show-costs is enabled; baseline Cypress swing-yard case from TN-157.
cable_highlead_tn147	tn147 (Madill 009 highlead)	5.5 m³ turns, 138 m average reach, ≈2.1 pcs/turn, manual falling defaults tuned to TN-147	grapple_yarder_madill009 CPI-aware rate; CLI output also prints the 1989 $/m³ cost column from TN-147.
cable_running_tr122_extended / _shelterwood / _clearcut	tr122-extended | tr122-shelterwood | tr122-clearcut	2.04/2.18/1.45 m³ turns, 104 m / 97 m / 106 m reach, observed pieces-per-cycle (2.2/3.2/2.6)	Uses the generic grapple_yarder cost role; the shelterwood preset now also applies the SR109 shelterwood penalty (volume ×0.495, cost ×1.38) via the partial-cut profile, while the extended/clearcut entries remain unchanged. CLI prints the TR-122 yarder/loader labour breakdown so you still see the published $/m³ numbers.
cable_running_sr54	sr54 (Washington 118A running skyline regression)	3.1 m³ turns, 82 m reach, ≈2.9 pcs/turn fed by the feller-buncher preset	Falls back to the generic grapple_yarder rate; rely on SR-54’s CPI-adjusted output or override the machine role if you have a local rental.
cable_running_tr75_bunched / _handfelled	tr75-bunched (System 1) / tr75-handfelled (System 2)	1.29 m³ @ 71 m and 1.28 m³ @ 76.6 m, 2.18 vs. 1.41 pcs/turn (mechanical vs. hand-felled)	Generic grapple_yarder cost role; CLI still reports the 1987 $/m³ numbers from TR-75.
cable_running_adv5n28_clearcut / _shelterwood	adv5n28-clearcut | adv5n28-shelterwood	2.2 m³ @ 480 m and 1.8 m³ @ 330 m plus skyline/helicopter comparison rows	grapple_yarder_adv5n28 cost role; banner also shows projected skyline vs. helicopter savings.
cable_running_fncy12	fncy12-tmy45 standing-in for the Thunderbird TMY45 + Mini-Mak II intermediate-support study	3.6 logs/turn, 176 m shift productivity, Cat D8/Timberjack support ratios baked into telemetry	grapple_yarder_tmy45 CPI-aware rate (Cat D8 + Timberjack standby surcharges included); CLI warns whenever lateral pulls exceed the TN-258 tension envelope.
cable_salvage_grapple	tn157 (Cypress) defaults + ADV1N5 salvage cues	Same payload/distance as cable_running with salvage-mode telemetry tags	Uses grapple_yarder_cypress7280; CLI/telemetry also carry the salvage processing mode.

Manual falling overrides follow the study provenance: TN-147/TN-157/ADV5N28 systems default to Douglas-fir 52.5 cm, the Roberts Creek TR-122 presets drop to 40 cm, and TR-75 hand-felled keeps a 45 cm reference. Mechanically bunched options (SR-54, TR-75 System 1) switch the felling job to feller-buncher so no manual falling cost is injected unless you explicitly request it. All of the new IDs appear in the medium/large synthetic tier mixes so the helper defaults flow into generated scenarios (and their machine cost banners) automatically.

Shovel Logger (Hoe-Chucker) Productivity

Primary-transport hoe chuckers are modeled through the Sessions & Boston (2006) serpentine model:

• shovel_logger role uses estimate_shovel_logger_productivity_sessions2006 with inputs mirroring the paper (passes between roads, swing length, strip length, volume per hectare, swing times/payloads, and walking speeds). Invoke via fhops.dataset estimate-productivity --machine-role shovel_logger.

• Key flags: --shovel-passes (number of swings), --shovel-swing-length, --shovel-strip-length, --shovel-volume-per-ha, and the swing/velocity parameters. Defaults follow Table 1 of the paper (16.15 m swing, 375 m³/ha, 20–30 s swing times, 0.7 kph travel speeds, 50 productive minutes/hour).

• FPInnovations TN-261 data introduced slope/bunching multipliers: use --shovel-slope-class to apply downhill (×1.1), level (×1.0), or uphill (×0.9) adjustments, and --shovel-bunching to pick between feller-bunched (×1.0) or hand-scattered (×0.6) stems. Stack an optional --shovel-productivity-multiplier for site-specific tweaks.

The helper reports cycle minutes, payload per cycle, and m³/PMH0 so you can test alternate swing counts or strip lengths. Hook it into harvest-system templates by assigning the shovel_logger job and, if needed, adding productivity_overrides for site-specific swing counts. Templates such as ground_hand_shovel and ground_fb_shovel already ship with overrides so the CLI automatically pulls their parameters when you pass --harvest-system-id or --dataset/--block-id.

Skyline Productivity Models

Use fhops.dataset estimate-skyline-productivity when the block’s harvest system references skyline_yarder jobs. The command now accepts --harvest-system-id, --dataset, and --block-id so defaults from the registry (e.g., cable_standing or cable_running) or the current scenario flow directly into the regressions. Available models:

• lee-uphill / lee-downhill – HAM300 case study (Lee et al. 2018, South Korea). Uphill only needs --slope-distance-m (calibrated 5–130 m); downhill also needs --lateral-distance-m (0–25 m) and --large-end-diameter-cm (~34 cm). The CLI emits a warning because these are short-span non-BC studies.

• tr125-single-span / tr125-multi-span – FPInnovations TR-125 standing skyline (Skylead C40 with Mini-Maki). Provide slope distance (10–420 m) and lateral distance (0–50 m). Payload defaults to 1.6 m³ (≈3.4 logs × 0.47 m³/log) but can be overridden; the CLI also prints the underlying cycle minutes so you can see the intermediate-support penalty. Pair these presets with inspect-machine --machine-role grapple_yarder_skyleadc40 (FERIC TN-201 cost entry) when modelling Skylead corridors. FNCY12/TN-258 Thunderbird spans should reference grapple_yarder_tmy45—its operating column now bundles the LeDoux (1984) yarder cost plus the FNCY12-derived support-machine surcharges (0.3335 SMH Cat D8 backspar + 0.2415 SMH Timberjack 450 trail support per productive yarder hour, see TN-258 / FNCY12 support utilisation) so --show-costs reflects the extra gear those corridors require. Whenever lateral pulls exceed ~30 m or skyline tension approaches the 140 kN ceiling documented in TN-258, estimate-skyline-productivity now prints a TN-258 warning and logs the skyline/guyline ratios (data lives in data/reference/fpinnovations/fncy12_tmy45_mini_mak.json) so analysts can flag hang-up risks alongside the costing output.

• fncy12-tmy45 – Thunderbird TMY45 + Mini-Mak II intermediate-support skyline (FNCY-12 / TN-258). The helper reports the observed shift output (default = steady-state months) divided by the 10 h shift to give m³/PMH and records the Cat D8 / Timberjack support ratios derived from FNCY12 Tables 2–3 (see TN-258 / FNCY12 support utilisation for the full derivation). Use --fncy12-variant to swap between overall, steady_state, or steady_state_no_fire averages, and keep --lateral-distance-m at or below 30 m if you want to stay inside the measured tension envelope—lateral pulls beyond that trigger the TN-258 warning automatically and mark the telemetry row so downstream costing knows hang-ups are likely. The dataset’s support_monthly_summary enumerates July (learning/no supports) versus the August–October support months with yarder SMH and the inferred Cat D8/Timberjack SMH, giving analysts a reproducible “supports vs. no supports” split even though TN-258 never published stopwatch utilisation.

• ledoux-skagit-shotgun / ledoux-skagit-highlead / ledoux-washington-208e / ledoux-tmy45 – LeDoux (1984) residue yarding regressions (Willamette & Mt. Hood studies). Supply slope distance plus --merchantable-logs-per-turn, --merchantable-volume-m3, --residue-pieces-per-turn, and --residue-volume-m3. Outputs report total payload, cycle minutes, and CPI-adjusted 1984 USD cost references (see inspect-machine --machine-role grapple_yarder_skagit_shotgun|_highlead|_washington_208e|_tmy45_residue). These models represent US residue trials—treat as comparative benchmarks rather than BC defaults.

• micro-master – Model 9 Micro Master yarder regression (FERIC TN-54, 1982). Defaults assume 3.2 pieces per turn (0.46 m³/piece) and the observed 5.96 min cycle at the 70 m spans logged on Vancouver Island. Supply --slope-distance-m (and optionally --pieces-per-cycle, --piece-volume-m3, or --payload-m3) to see the computed payload, cycle minutes, and productivity; useful for compact skyline/thinning setups where Madill/Cypress presets are overkill.

• tr127-block1 … tr127-block6 – FPInnovations TR-127 block-specific tower-yarders. Blocks 1 & 3 require two lateral inputs (--lateral-distance-m plus --lateral-distance-2-m); Blocks 5 & 6 require --num-logs. Range checks match the publication and the table output shows cycle minutes alongside m³/PMH.

• mcneel-running – McNeel (2000) Madill 046 longline running skyline. Requires horizontal span (--horizontal-distance-m or reuses --slope-distance-m), deflection (--vertical-distance-m), lateral distance, and optional --pieces-per-cycle / --piece-volume-m3 plus --running-yarder-variant (Yarder A/B). Selecting --harvest-system-id cable_running pre-populates those inputs automatically.

• aubuchon-standing – Hensel et al. (1979) Wyssen standing skyline (Aubuchon 1982 Appendix A Eq. 15). Provide --logs-per-turn, --average-log-volume-m3, --crew-size, slope, and lateral distance. Calibrated for 45–75 % corridors with 3.5–6 logs/turn; the CLI warns that it is a WA/ID dataset.

• aubuchon-kramer – Kramer (1978) standing skyline with carriage-height and chord-slope predictors. Supply --carriage-height-m and --chordslope-percent in addition to the Wyssen-style inputs; --harvest-system-id cable_standing applies 3.8 logs/turn, 0.45 m³/log, and −15 % chord slope defaults for coastal chokers. The trials only observed chord slopes between −22 % (downhill) and +3 % (slight uphill), so the helper now warns/flags telemetry when you stray outside that envelope.

• aubuchon-kellogg – Kellogg (1976) tower-yarder regression (Oregon) with lead-angle and choker-count predictors. Provide --lead-angle-deg and --chokers with the usual log/volume inputs; payloads are converted to cubic feet internally to match the source, and the CLI warns that it’s not BC data. Lead angles were limited to ±90° and the study only used 1–2 chokers per turn, so the helper now prints/telemeters calibration warnings whenever those values drift.

Helicopter longline work still lives under fhops.dataset estimate-productivity --machine-role helicopter_longline. Those helpers now read from data/productivity/helicopter_fpinnovations.json (ADV3/4/5/6 series) so each model has a reference preset plus a matching machine-rate role:

• --helicopter-model accepts lama, kmax, bell214b, ka32a (Kamov KA-32A pole logging), or s64e_aircrane. When --helicopter-preset is omitted, the CLI applies the default preset for that model (marked with ★ in fhops.dataset helicopter-fpinnovations); specify a preset ID (e.g., s64e_grapple_retention_adv5n13) to seed the flight distance, payload, load-factor, weight→volume, and delay minutes directly. The CLI and telemetry banner call out whichever preset supplied the defaults so downstream costing can audit the assumption.

• fhops.dataset helicopter-fpinnovations lists every preset, the observed m³/shift, turn times, and CPI metadata. Add --operation-id <preset> to view the full cycle breakdown and scan notes without reopening the PDFs.

• --show-costs now hooks into the CPI-aware helicopter machine-rate entries (helicopter_lama/_kmax/_bell214b/_ka32a/_s64e_aircrane) so heavy/medium/light lift cost banners cite the FPInnovations Appendix-II splits directly. inspect-machine --machine-role helicopter_s64e_aircrane (etc.) surfaces the same rates outside the productivity helper.

Selecting --harvest-system-id helicopter still applies the Bell 214B defaults, but the new preset plumbing means Mission-style Aircrane or KA-32A pole-logging runs can be reproduced without hand-entering the published flight distance/payload values.

Every skyline model prints the assumed payload and m³/PMH0 result; the McNeel and Aubuchon helpers also show delay-free cycle minutes so you can see the impact of deflection, carriage height, lead angle, or extra chokers. Telemetry rows capture all skyline predictors (horizontal/vertical distance, pieces, piece volume, carriage height, chord slope, lead angle, chokers, running-yarder variant) so harvest-system overrides are traceable. Append --show-costs to print the CPI-adjusted machine-rate summary for any preset that has a matching entry in the Skyline cost matrix below.

TR-127 predictor ranges

Use the table below when selecting the block-specific TR-127 presets. Blocks differ in their required inputs and the calibrated ranges of those inputs (from Appendix VII of TR-127). Values outside these ranges still compute but the CLI emits calibration warnings and telemetry flags the deviation.

TR-127 block summary

Block	Scenario	Required predictors	Calibrated range(s)
1	Cedar partial cut (two lateral pulls)	latd (primary lateral), latd2 (secondary lateral)	latd 0–8 m; latd2 0–8 m
2	Spruce clearcut	sd (slope distance)	sd 225–770 m
3	Steep partial cut	latd (lateral distance)	latd 0–65 m
4	Small lateral, coastal block	latd (lateral distance)	latd 0–40 m
5	Mixed partial cut (IFN block)	sd (slope distance), latd (lateral), logs (logs/turn)	sd 70–650 m; latd 0–80 m; logs 1–10
6	Long-span partial cut (north block)	sd (slope distance), latd (lateral), logs (logs/turn)	sd 265–716 m; latd 0–75 m; logs 1–7

Skyline cost matrix

When you need CPI-adjusted hourly costs, run fhops.dataset inspect-machine --machine-role <role> (or reference data/machine_rates.json) using the role associated with your preset. The table below summarizes which machine-rate entry lines up with each skyline helper. Presets without BC cost coverage are flagged so you know to supply your own rates.

Skyline preset → machine-rate mapping

Preset(s)	Source/system	Machine-rate role	Notes
tr125-single-span / tr125-multi-span / tr127-block1–block6	FPInnovations TR-125/127 Skylead C40 corridors	grapple_yarder_skyleadc40	1991 CAD rate (FERIC TN-201). Reuse this CPI-aware entry for both single- and multi-span corridors until dedicated TR-127 costs surface.
fncy12-tmy45	Thunderbird TMY45 + Mini-Mak II intermediate supports (FNCY12/TN258)	grapple_yarder_tmy45	1992 CAD LeDoux-derived rate with Cat D8/Timberjack standby baked in (0.3335/0.2415 SMH; see TN-258 / FNCY12 support utilisation). Telemetry logs the support ratios automatically.
mcneel-running	McNeel (2000) Madill 046 running skyline	grapple_yarder_cypress7280	Uses the TN-157 Cypress swing-yard rate (1991 CAD) for long-span running skyline conversions.
tn173-ecologger / tn173-gabriel / tn173-christie / tn173-teletransporteur / tn173-timbermaster-1984 / tn173-timbermaster-1985	TN-173 compact skyline fleet (Eastern Canada)	skyline_ecologger_tn173 / skyline_gabriel_tn173 / skyline_christie_tn173 / skyline_teletransporteur_tn173 / skyline_timbermaster_tn173	Each preset has a dedicated 1991 CAD entry mirroring the original FERIC case study; CPI helper inflates to 2024.
hi-skid	FNG73 Hi-Skid short-yarding truck	skyline_hi_skid	1999 CAD attachment + operator (ADV2N62 wage). inspect-machine reports the CPI-adjusted rate.
tn147	TN-147 Madill 009 highlead skyline (coastal BC)	grapple_yarder_madill009	1991 CAD owning/operating split captured from the FERIC report; used whenever the TN147 helper runs.
tn157	TN-157 Cypress 7280B swing yarder + UH14 trail spar	grapple_yarder_cypress7280	CPI-aware rate includes the trail-spar allowances; also reused by mcneel-running when modelling Cypress conversions.
adv5n28-clearcut / adv5n28-shelterwood	ADV5N28 Madill 071 skyline conversions (helicopter replacement)	grapple_yarder_adv5n28	2002 CAD Advantage owning/operating split with CPI adjustments; telemetry logs the skyline-vs-helicopter savings.
ledoux-skagit-shotgun / ledoux-skagit-highlead / ledoux-washington-208e / ledoux-tmy45	LeDoux (1984) residue skyline trials (Skagit BU-94, Washington 208E, Thunderbird TMY-45)	grapple_yarder_skagit_shotgun / grapple_yarder_skagit_highlead / grapple_yarder_washington_208e / grapple_yarder_tmy45_residue	1984 USD owning/operating splits converted via the CPI helper for the residue presets.
Presets without dedicated BC rates (lee-uphill, lee-downhill, aubuchon-*, micro-master)	South Korea / US Pacific Northwest studies outside the FPInnovations catalog	—	Supply your own rate (or reuse grapple_yarder generic) when running costing workflows; these remain non-BC references.

TN-258 / FNCY12 support utilisation

The Thunderbird TMY45 intermediate-support study (FNCY12 + TN-258) is the only FPInnovations source that publishes both the month-by-month yarder SMH and the crew/man-day split needed to back-calculate support-machine utilisation. We take the 2.5 man-day delta between the Thunderbird and Skylead crews (Table 3), scale it by the 23 % of the harvest area that required intermediate supports (Table 2), then split the resulting labour tax between the Cat D8 backspar and Timberjack 450 trail-support duties using the TN-157 backspar-share statistics (≈58 % backspar, ≈42 % trail support). The helper script scripts/fncy12_support_split.py reproduces the math from the structured dataset at data/reference/fpinnovations/fncy12_tmy45_mini_mak.json so costing workflows can be audited end-to-end.

Support ratios per productive yarder SMH

Preset / helper	Cat D8 backspar	Timberjack 450 trail support	Notes / source
cable_running_fncy12 / fncy12-tmy45 / grapple_yarder_tmy45	0.3335	0.2415	Derived from FNCY12 Tables 2–3 plus TN-157 backspar-share stats; see data/reference/fpinnovations/fncy12_tmy45_mini_mak.json or rerun scripts/fncy12_support_split.py for the full calculation.

If future FPInnovations payroll or utilisation logs surface (e.g., Jay-Vee/Bell Pole internal records) we can swap in the published minutes, but until then these ratios represent the best-available data and replace the older TN-157 proxy (0.25/0.14) throughout the skyline cost banner, telemetry, and harvest-system overrides.

Partial-cut profile registry

fhops.dataset estimate-skyline now understands --partial-cut-profile <id> (and the skyline harvest systems listed earlier set the appropriate ID for you). Profiles live in data/reference/partial_cut_profiles.json and bundle the observed volume/cost multipliers from FPInnovations partial-cut trials so productivity, costing banners, and telemetry all flag the penalty without reopening the PDFs.

Available partial-cut profiles

Profile ID	Source / description	Volume multiplier	Cost multiplier
sr109_patch_cut / sr109_green_tree / sr109_shelterwood	SR-109 MASS patch, green-tree, and shelterwood forwarding trials (excavator + tracked skidder relays)	0.846 / 0.715 / 0.495	1.10 / 1.10 / 1.38
adv11n17_trial1 / trial2 / trial3	Advantage 11 No. 17 partial-cut trials (motor-manual CTL, cable skidder full-tree, mechanised full-tree)	0.775 / 0.833 / 0.925	1.52 / 1.20 / 1.21
adv1n22_group_selection	Advantage 1 No. 22 group-selection opening (Barkerville Corridor small-opening program)	0.918	1.06
tn199_partial_entry	Quadra Island JD450G crawler winching through dispersed retention (12–15 % removal per entry)	0.177	2.14
adv9n4_interface_thinning	Alex Fraser Research Forest interface fuel-reduction thinning (tractor/skidder/ATV skidding + manual piling/burning)	0.012	27.0

Profiles with neutral multipliers (e.g., adv11n17_trial4) act as baselines and are omitted for brevity. Update the JSON file and re-run the CLI when new FPInnovations partial-cut datasets are digitised—the skyline helper picks up additions automatically.

Road & Subgrade Construction References

Roadbuilding presets are still in flight, but the foundational FPInnovations dataset is now captured in data/reference/fpinnovations/tr28_subgrade_machines.json (FERIC TR-28 “Productivity and Cost of Four Subgrade Construction Machines”). Each machine entry includes movement costs, phase-level cycle rates (logging, stumping, clearing, excavation), utilization, per-station cost, and roughness indicators so the upcoming road-cost helper can quote realistic BC numbers without reopening the PDF.

• Cat 235 hydraulic backhoe – 3.15 stations/shift (≈96 m), 5.56 CAD/m subgrade cost, 0.63 min working cycle, and separate mineral vs. overburden excavation rates (174 vs. 127 m³/h) for cut/fill planning.

• Cat D8H bulldozer – 2.61 stations/shift (≈80 m), 6.37 CAD/m, and higher walking speed (10.5 km/h) that will feed the cut-and-fill vs. push distance toggles; logging/stumping cost per stem already stored for landing-clearing presets.

• American 750C line shovel – 3.07 stations/shift, 6.13 CAD/m with lower utilization (79 %) and steeper movement costs (two trucks + 4 h lowbed) to cover dipper-shovel mobilization cases.

• Poclain HC300 hydraulic shovel – 1.78 stations/shift, 10.8 CAD/m, high excavation time (47 h) and the worst roughness indicator (4.02 m²/100 m), making it the default cautionary example for wet-site subgrade finishing.

Use these entries when drafting the road-cost helper schema (owning/operating split, movement surcharge, quality index) and tie the soil-protection warnings back to TR-28’s roughness indicators plus ADV4N7/FNRB3 once those documents are staged. Until the helper ships, cite this section in planning docs whenever roadbuilding data is required so we avoid redundant TR-28 rescans.

For a quick terminal summary, run fhops dataset tr28-subgrade. The command reads the structured JSON, lets you filter by role (e.g., bulldozer vs. backhoe), sort by unit cost/stations/roughness, and prints the CPI-base metadata so road cost planning sessions can pull the TR-28 numbers without reopening the PDF.

Need a CPI-adjusted estimate for a specific spur? Run fhops.dataset estimate-road-cost --machine <slug> --road-length-m <m>. Pick any machine slug from the tr28-subgrade table (for example caterpillar_235_hydraulic_backhoe) and the CLI will apply the 1976 CAD unit cost to your requested length, inflate it to the current StatCan CPI target, and (optionally) add the published mobilisation cost (toggle with --exclude-mobilisation). Output lists base-year + 2024 CAD totals alongside the implied number of stations (30.48 m each) and estimated shifts so you can sanity-check the assumptions before wiring the numbers into a planning sheet. When a scenario carries a road_construction table (id, TR-28 slug, length, mobilisation flag, soil profile IDs), fhops.dataset estimate-cost --dataset … will pull those defaults automatically (use --road-job-id if multiple entries exist) and print the structured FNRB3/ADV4N7 soil profiles defined in data/reference/soil_protection_profiles.json so planners see exactly which compaction or ground-pressure constraints were assumed. Skyline/cable harvest-system templates now push those road defaults for you: any block that references a preset listed in fhops.scheduling.systems.SYSTEM_ROAD_DEFAULTS causes the scenario (and the synthetic dataset bundles) to emit the matching road_construction row, so estimate-cost and telemetry immediately reflect the paired TR-28 machine, length, mobilisation flag, and soil-profile IDs without touching the CSVs.

ADV4N7 and ADV15N3 inputs now drive automatic support penalties when those soil profiles are present. The default skyline road entries include adv4n7_compaction so fhops.dataset estimate-cost multiplies the TR-28 unit cost by 1.15 (“some” risk: reduce loads and limit passes to ≤3) and logs the source in telemetry. Select --road-compaction-risk low|some|high when you need to override the default, or remove the soil profile entirely if the block is on dry/frozen ground. Road jobs that reference caterpillar_d8h_bulldozer now pull the ADV15N3 tractor-efficiency study: whenever compaction risk is some or high the CLI applies the Cat D7R low-speed penalty (≈1.16× fuel / SMH) and prints a “[Support penalty applied]” banner citing ADV4N7 + ADV15N3. Use --road-tractor-drive if you want to model the hydrostatic (John Deere 950J) or diesel-electric (Cat D7E) drives from ADV15N3; the helper applies the published fuel-intensity delta even when soils are firm and still honours the low-speed penalty whenever the ADV4N7 risk is “some” or “high”. Telemetry captures the compaction/tractor multipliers so downstream costing can see exactly why a scenario’s road add-on changed.

Handfalling Cost Reference

Manual falling productivity/costs now live in data/reference/fpinnovations/tn98_handfalling.json (FERIC TN-98, Peterson 1987). Use fhops dataset tn98-handfalling --species douglas_fir --dbh-cm 32.5 to pull the regression-based cutting time, interpolated limbing delay, and cost-per-tree / cost-per-m³ for a given DBH class (defaults to the study’s “all species” regression). Add --show-table when you want to review the observed per-diameter rows before copying numbers into skyline costing sheets. fhops.dataset estimate-skyline-productivity now accepts --manual-falling with optional --manual-falling-species and --manual-falling-dbh-cm so TN-98 falling costs show up beside the skyline output. Harvest systems that include a hand-faller job auto-enable those settings (e.g., cable_micro_* presets default to hemlock @ 32.5 cm, cable_running presets default to Douglas-fir @ 52.5 cm) and print a reminder when the defaults kick in.

Use fhops dataset tn82-ft180 to compare the FMC FT-180 vs. John Deere 550 tracked skidders from TN-82 (clearcut vs. winter right-of-way). The command prints m³/PMH and per-shift stats for each site, making it easy to benchmark steep-slope ground alternatives alongside the skyline helpers.

Use fhops dataset adv6n25-helicopters to review the ADV6N25 light-lift operation (Eurocopter Lama + K-1200 K-Max). The CLI prints productivity/cost rows for each helicopter and, with --show-alternatives, the “in-woods manufacturing” and “single-pass K-Max” cost scenarios discussed in the report.

CTL Harvester Productivity Models

fhops.productivity.harvester_ctl hosts the shortwood harvester regressions. The inaugural entry is the ADV6N10 single-grip model:

• adv6n10 – Gingras & Favreau (2005) boreal multi-product study. Use when operators sort several products per pass and need visibility into harvester-side penalties. Required CLI flags: --ctl-stem-volume (m³/stem), --ctl-products-count (number of products sorted per cycle), --ctl-stems-per-cycle, and --ctl-mean-log-length (m). Invoke via fhops dataset estimate-productivity --machine-role ctl_harvester to see the ADV6N10 result.

• adv5n30 – Meek (2004) ADV5N30 Alberta white-spruce thinning. Provides removal-level productivity (30/50/70 %) plus an optional --ctl-brushed flag (adds the observed 21 % boost when brush crews pre-clear). Supply --ctl-removal-fraction (0–1 within the 0.30–0.70 window) to interpolate between the published removal levels.

• tn292 – Bulley (1999) TN292 Alberta thinning (Kenmatt/Brinkman). Models harvester productivity versus tree size and stand density. Requires --ctl-stem-volume (m³/stem), --ctl-density (trees/ha), and --ctl-density-basis (pre or post harvest) to match the published regressions.

• kellogg1994 – Kellogg & Bettinger (1994) western Oregon CTL thinning case (Timberjack 2518 harvester). Supply --ctl-dbh-cm (study range 10–50 cm). The helper outputs delay-free m³/PMH so you can compare harvester productivity directly with the paired Kellogg forwarder models.