Update
This commit is contained in:
46
verification/SAAB_GT17.md
Normal file
46
verification/SAAB_GT17.md
Normal file
@@ -0,0 +1,46 @@
|
||||
# Saab B205 with GT17 and ECU boost control
|
||||
|
||||
The desktop Saab menu entry and `--scenario b205` now select an estimated GT1752
|
||||
turbo configuration. `--turbo gt17` makes that choice explicit; `--turbo none`
|
||||
selects the previous naturally aspirated baseline. CLI and WAV capture follow
|
||||
the same defaults. `SaabB205Scenario` preserves its baseline API;
|
||||
`SaabTurboScenario` is the boosted convenience class.
|
||||
|
||||
The GT17 calibration uses a 26 mm compressor tip radius, 19 mm effective turbine
|
||||
radius, 2e-5 kg m² rotor inertia, 420/350 mm² effective compressor/turbine areas,
|
||||
70% efficiencies, 100 cc housing volumes and a 24 mm wastegate. These are model
|
||||
estimates, not a measured Saab map or verified rotor inertia. The Garrett catalog
|
||||
identifies the Saab GT1752 application:
|
||||
https://www.turbomaster.info/eng/downloads/catalog_garrett.pdf
|
||||
|
||||
The ECU requests up to 45 kPa gauge boost, ramping with pedal and RPM. A PI
|
||||
controller uses manifold pressure filtered over 50 ms, conditional integration,
|
||||
and bounded pneumatic bleed duty (0–85%). Duty reduces the pressure signal to
|
||||
the existing spring/diaphragm actuator. Actual valve travel remains damped and
|
||||
boost emerges from the physical rotor, restrictions and conserved gas transfers.
|
||||
Zero duty restores the mechanical pressure-can response; it does not forcibly
|
||||
open the valve or impose a manifold pressure. Targets below spring boost cannot
|
||||
be achieved by wastegate bleed alone.
|
||||
|
||||
Lift-off, fuel cut and invalid sensor data request zero duty and clear the PI
|
||||
integral. Invalid boost sensors cut fuel. Manifold or charge pressure above
|
||||
75 kPa gauge triggers fuel cut, recovering below 55 kPa gauge. This is a basic
|
||||
estimated safety calibration, not a reproduction of Saab Trionic software.
|
||||
No intercooler, compressor surge/BOV model or knock model has been added;
|
||||
charge heating and exhaust backpressure remain present.
|
||||
|
||||
The regression holds the engine at 3,500 RPM, applies full pedal for six seconds,
|
||||
then releases for three. It checks boost settling within 8 kPa of the 45 kPa
|
||||
target, shaft spool-up/down, finite audio, no pipe positivity repairs, no housing
|
||||
donor limiting, and gas/rotor energy residual below 0.1 J. Controller checks cover
|
||||
overboost, recovery, pedal release and an invalid sensor. The HUD shows target
|
||||
gauge boost, actuator bleed duty, turbo RPM and absolute gas pressures.
|
||||
|
||||
```sh
|
||||
cmake --build build-desktop --target fluidsim_desktop -j4
|
||||
./build-desktop/fluidsim_desktop --scenario b205 --turbo gt17
|
||||
./build-audit/fluidsim_scenario_benchmark 6 loaded b205-gt17
|
||||
```
|
||||
|
||||
All ten CTest suites pass, including the new ten-second Saab turbo regression.
|
||||
The GT17 benchmark holds 5,500 RPM, below the Saab's 6,500 RPM limiter.
|
||||
Reference in New Issue
Block a user