# Evidence and assumptions register

Generated from evidence.json (also shown in the website below the analytics). "Illustrative assumption" rows are study inputs chosen to make the model run, not findings. Project PDFs are cited by title and PDF page; they are not published with the site. The full screening of the 46 supplied PDFs is in literature_review/REVIEW.md in the project workspace. This register does not claim that every paper or regulation has been fully implemented.

## Driver hours (standard case)

- **Type:** regulation
- **Source and date:** [DVSA: goods vehicles, assimilated and AETR rules on drivers' hours](https://www.gov.uk/guidance/drivers-hours-goods-vehicles/1-assimilated-and-aetr-rules-on-drivers-hours) · page updated 1 Jul 2026; checked 28 Oct 2026
- **What it supports:** Break of at least 45 min after no more than 4.5 h driving (or 15 + 30 split); 9 h daily driving (10 h twice a week); regular daily rest of 11 h completed within 24 h of the previous rest; 56 h weekly and 90 h fortnightly driving; reduced/split daily rest and weekly-rest rules.
- **Model translation:** GB standard mode fixes 270 min driving / 45 min break, 540 min daily driving and 660 min regular daily rest started by 13 h after duty start. Weekly/fortnight caps use supplied prior minutes.
- **Limitations:** Supported subset only: no split breaks, 10 h extensions, reduced or split daily rest, weekly rest, other work, working-time rules, multi-manning, exemptions or tachograph records. No legal-compliance claim.

## Queue waiting is not a break

- **Type:** model rule
- **Source and date:** [Conservative modelling choice informed by the DVSA break definition](https://www.gov.uk/guidance/drivers-hours-goods-vehicles/1-assimilated-and-aetr-rules-on-drivers-hours) · 28 Oct 2026
- **What it supports:** A qualifying break must be uninterrupted rest; a queue can require the driver to move the vehicle.
- **Model translation:** Queue minutes never reset the driving clock. Rest needs a parking space or a connected bay and must complete in full.
- **Limitations:** Some real queues may allow genuine rest; modelling that would need an explicit no-movement queue rule.

## Queue management and alternative sites

- **Type:** evidence
- **Source and date:** BSI Flex 2071 v1.0:2024-06 (project folder, R02) PDF p14, §5.1.2–5.1.3 · 2024-06
- **What it supports:** Availability information, pre-booked versus ad-hoc access, queue management and information about alternative charge points.
- **Model translation:** Three comparable queue policies, traceable allocation records and bounded diversion to alternatives.
- **Limitations:** Not certified compliance; no booking, live availability or reservation enforcement.

## Bays occupied by breaks; reservations are not availability

- **Type:** evidence
- **Source and date:** Electric Freightway report 2 (R36) PDF pp26–27, 33; FLEXMCS guidance (R28) pp14, 32 · as supplied
- **What it supports:** Bays blocked by driver breaks, booking delays, unused reserved bays, monitoring and availability displays.
- **Model translation:** Charging completion, bay release and departure are separate; a resting driver holds the bay unless an authorised mover and parking exist. Diversion forecasts are labelled as not reservations.
- **Limitations:** No no-show, cancellation or booking model.

## Occupied bay versus active charging

- **Type:** evidence
- **Source and date:** [Fraunhofer HoLa interactive tool (public methodology)](https://hochleistungsladen-lkw.de/hola-en/results/interactive-online-tool.php) · reviewed in project WEB_LINK_REVIEW
- **What it supports:** Distinguishes parked/occupied bays from charging power; stop purposes; break versus demand charging; waiting averages over affected users.
- **Model translation:** Separate occupied-bay and charging-bay utilisation; holding state at zero power; started-session wait percentiles reported separately from trucks still queued.
- **Limitations:** German reference-site assumptions; no HoLa numbers were reproduced or imported.

## Priority beyond SOC

- **Type:** evidence
- **Source and date:** CSMS2023-0003 parking-lot multi-agent EV scheduling (R07) PDF pp4–5 · 2023
- **What it supports:** Charging urgency can combine energy need and departure constraints, not SOC alone.
- **Model translation:** Research policy: next-leg energy risk → waiting-age protection → SOC → FIFO. Battery age never grants priority.
- **Limitations:** Car parking-lot context; the HGV ranking is a research hypothesis to compare against FIFO, not a published standard.

## Waiting-time thresholds

- **Type:** evidence
- **Source and date:** T&E 2022 grid integration of long-haul truck charging (R18) PDF pp84–86 · 2022-01
- **What it supports:** A stochastic queue model with a study-specific 15-minute adequacy threshold and no alternative station.
- **Model translation:** Reports mean, p50, p95 and maximum wait plus trucks still queued; histogram bins are descriptive only.
- **Limitations:** 15 minutes is not adopted as a service standard.

## 44 t consumption, 2025 vs 2030/2035

- **Type:** evidence
- **Source and date:** Element Energy, 'Why GB's long-distance HGVs can go battery electric' (file 20230425_TE_BEHGV_public_charging_report.pdf) PDF p60 · 2023-04
- **What it supports:** Illustrative 44 t vehicle: 1.74 kWh/km (2025) and 1.58 kWh/km (2030/2035) including charger efficiency of 86% (2025) and 95% (2030/2035).
- **Model translation:** Presets use 1.5 kWh/km battery side for 44 t classes in all years (1.74 × 0.86 ≈ 1.58 × 0.95 ≈ 1.50) and hub efficiency 0.86 (2025) / 0.95 (2030, 2035).
- **Limitations:** Continuous 44 t on a motorway/dual-carriageway mix; payload, gradient, weather and HVAC are not modelled. Regional rigid 1.0 kWh/km is illustrative.

## Charger power for 2030 sites

- **Type:** evidence
- **Source and date:** Element Energy (same report) PDF pp44–45 · 2023-04
- **What it supports:** 2030 network mixes 350 kW and 1 MW charging; a 2030 site typically needs 1–3 MW chargers.
- **Model translation:** Hub preset 2025: 350 kW CCS; 2030/2035: 1,000 kW CCS|MCS. Charger count and grid limit are not changed by the preset.
- **Limitations:** National network analysis, not a site design. Real per-connector limits and shared cabinets are not modelled.

## MCS connector envelope

- **Type:** evidence
- **Source and date:** [CharIN Megawatt Charging System](https://www.charin.global/technology/mcs/) · reviewed in project WEB_LINK_REVIEW
- **What it supports:** MCS technical envelope (1,250 V, 3,000 A) and connector compatibility as a separate question from delivered power.
- **Model translation:** Connector sets on hubs and vehicles; compatibility = shared standard. Delivered power = min(charger, vehicle × derating, shared grid budget).
- **Limitations:** The envelope is not a vehicle charging curve; no taper is modelled.

## Charging power versus SOC (taper)

- **Type:** illustrative assumption
- **Source and date:** Study assumption; project WEB_LINK_REVIEW notes that an MCS envelope is not a vehicle charging curve · 28 Oct 2026
- **What it supports:** Delivered power falls at high SOC, so constant-power models overstate late-session power and understate bay time.
- **Model translation:** P = acceptance × derating × f(SOC), f piecewise linear: 1.0 to 60% SOC, 0.7 at 80%, 0.4 at 90%, 0.15 at 100%, evaluated at the start of each minute. Per-truck curves override it.
- **Limitations:** Shape is not calibrated to any vehicle or chemistry; temperature and pack-level limits not modelled. Replace with OEM or session-log curves.

## Uncertainty from random arrivals

- **Type:** engineering method
- **Source and date:** Independent replications with common random numbers (standard discrete-event simulation practice) · —
- **What it supports:** A single stochastic run is one sample; decisions need the expected value and its spread.
- **Model translation:** Repeated runs with seeds base…base+n−1; mean with t-based 95% CI, 5th–95th percentile of runs, and paired case − current differences per seed.
- **Limitations:** Covers arrival randomness only (route, class, SOC, SOH, car arrivals); not parameter uncertainty in consumption, curves or stop rates unless those are varied as scenarios.

## Battery capacity and DC acceptance by year

- **Type:** illustrative assumption
- **Source and date:** Study assumption (no supplied source calibrates it) · 28 Oct 2026
- **What it supports:** —
- **Model translation:** 2025: 300/540/600 kWh at 250/400/400 kW; 2030: 400/650/750 kWh at 400/750/1,000 kW; 2035: 450/750/900 kWh at 400/1,000/1,000 kW.
- **Limitations:** Not a forecast. Replace with OEM specifications or fleet data before drawing conclusions.

## Route planning around infeasible legs

- **Type:** engineering method
- **Source and date:** Shortest-path search (Dijkstra) with feasibility constraints; motivated by Electric Freightway / BETT range and charging-stop planning (R37, R29) · 28 Oct 2026
- **What it supports:** Operators plan charging stops when a leg exceeds vehicle range; a blocked-only model understates network usefulness.
- **Model translation:** Fewest travel minutes over supplied links; each leg drivable from full with reserve and within a fresh driving allowance; intermediate hubs need a compatible charger; ≤ 6 legs; applied only when the next leg is infeasible.
- **Limitations:** Minimises driving time, not total time including charging and queues; no road routing, traffic or charger availability forecast.

## Arrival process

- **Type:** engineering method
- **Source and date:** Non-homogeneous Poisson thinning (Lewis & Shedler, 1979); Poisson arrivals as a modelling assumption in R20 pp24–25 · —
- **What it supports:** Reproducible random arrivals with a time-varying rate.
- **Model translation:** rate(t) = base rate × clock-hour multiplier; Python random.Random(seed); spec stored in the study JSON.
- **Limitations:** Profile shapes are illustrative. Calibrate with DfT raw counts by hour and direction; AADF is road flow, not hub arrivals.

## Traffic data for calibration

- **Type:** data source
- **Source and date:** [DfT road traffic statistics (count points, raw counts, API)](https://roadtraffic.dft.gov.uk/downloads) · reviewed in project WEB_LINK_REVIEW
- **What it supports:** Direction-specific hourly HGV counts at count points.
- **Model translation:** Not yet imported. Intended input for the hourly arrival profile.
- **Limitations:** Missing hours are not zeros; share of HGVs stopping at a hub must be assumed or observed.

## Grid capacity per hub

- **Type:** project requirement
- **Source and date:** Project brief (supervisor requirement) · Oct 2026
- **What it supports:** Each hub must have at least 3,000 kW available supply.
- **Model translation:** Validation rejects < 3,000 kW. This is available capacity; demand follows chargers, vehicles and arrivals.
- **Limitations:** Not a network-headroom assessment.

## Power sharing at a hub

- **Type:** engineering method
- **Source and date:** Max-min fair allocation (hgv_model/simulation.py, tested) · —
- **What it supports:** Shared grid budget split fairly, with unused share redistributed from vehicles at their cap.
- **Model translation:** Budget = (grid limit − background load) × efficiency; per-vehicle cap = min(charger, acceptance × factor, remaining energy per minute).
- **Limitations:** No charging curve, phase imbalance, losses beyond fixed efficiency, or smart tariff scheduling.

## Battery health

- **Type:** engineering method
- **Source and date:** Model definition · —
- **What it supports:** —
- **Model translation:** Usable capacity = nominal usable × SOH, applied once; SOC is relative to that capacity; calendar age is metadata only.
- **Limitations:** No degradation law; would need chemistry, temperature and duty-cycle calibration.

## Shared charging with cars

- **Type:** model rule
- **Source and date:** eFreight 'The Road Ahead' (R46): shared/depot access context · as supplied
- **What it supports:** Shared access to HGV-capable infrastructure is a recognised option.
- **Model translation:** Only designated shared bays; car starts only if no HGV waits and the HGV guard of free bays remains; non-preemptive; car leaves on completion or after its maximum wait.
- **Limitations:** Car arrival rates and energy are illustrative; no payment, booking or rental.

## Grid impact boundary

- **Type:** engineering method
- **Source and date:** [IEA EV charging and grid integration tool technical note PDF pp5–7, 14, 22, 28–29](https://www.iea.org/data-and-statistics/data-tools/electric-vehicle-charging-and-grid-integration-tool) · reviewed in project WEB_LINK_REVIEW
- **What it supports:** Demand-profile tools do not model network power flow.
- **Model translation:** Outputs are charging demand profiles and headroom against a supplied limit, exportable for a solver.
- **Limitations:** Voltage, feeder loading and losses need pandapower/OpenDSS or similar; not connected yet.

## Depot ratios are case-specific

- **Type:** evidence
- **Source and date:** Guidelines: charging infrastructure for truck depots (R03); depot fleet charging paper (R15) · as supplied
- **What it supports:** Vehicle-to-charger ratios depend on schedules and should not be universal defaults.
- **Model translation:** No ratio is assumed; charger counts are explicit hub inputs compared by scenario.
- **Limitations:** —

## Road distances and times

- **Type:** illustrative assumption
- **Source and date:** Example study inputs · —
- **What it supports:** —
- **Model translation:** Directed links supply km and minutes; nothing is inferred from coordinates.
- **Limitations:** Replace with a verified HGV route matrix; no traffic or speed-limit model.