New Mars Forums
You are not logged in.
- Topics: Active | Unanswered
Announcement
#276 2026-01-21 11:37:39
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
This run from zero to 4 was carried out in pursuit of the cause of incorrect thrust reported from initial tests of the NewNozzle case.
We have NO nozzle installed, but we were getting upwards of 30 force-tons of thrust ... OK ... we have an error in our configuration.
However, the ** nature ** of the error is not obvious, or it would have been spotted before the run was first launched.
The working theory in play is that a feature of OpenFOAM that provides automatic input of mass may be the source of the incorrect results.
The feature is nice... it allows the operator to specify a flow of mass into the model regardless of whatever else is going on. I've been using this feature from the beginning, since it was available and it definitely does what it is advertised to do. However, unbeknownst to me, this feature had the non-obvious side effect of adding energy to the situation without advertising. In the tail of the log of a run from 0 to 4, with a new report added, we can see that pressure at the intake is greater than the average pressure in the pipe. A possible explanation (not yet confirmed) is that the intake feature is quietly pushing new hydrogen into the heating pipe against the existing pressure, and in so doing helping to create thrust at the exit.
Another way of thinking about this is the energy budget... Our design calls for an energy budget of 40 MW. However, if the intake feature is doing what i think it is doing, it is adding some amount of energy to the budget in order to force hydrogen into a chamber that is already at a high level of pressure. Whatever that energy may be is not accounted for at present.
In a Real Universe engine, the energy required to pump propellant into a combustion chamber against the very high pressures already present comes from the propellant budget, except for the very few rockets that use electric (battery powered) pumps.
Here is the data from the end of a run from zero to 4 seconds, using the standard 2 kg/s feed and 40 MW of budgeted energy.
ExecutionTime = 11200.03 s ClockTime = 11207 s << 4 seconds in under 4 hours
volFieldValue p_avg write:
volAverage(heaterCells) of p = 720582.8 << Average pressure in all 9600 cellsvolFieldValue T_avg write:
volAverage(heaterCells) of T = 897.7824 << Average temperature in 9600 cellsvolFieldValue rho_mass write:
volIntegrate(heaterCells) of rho = 4.952744 << Total mass - most of initial load should be gone at 4 secondsvolFieldValue T_min_global write:
min() of T = 26.53625 at location (-239.5 0.175 2.292485e-17) in cell 8880 << Incoming gas is at 20 KelvinvolFieldValue T_max_global write:
max() of T = 2762.275 at location (4.5 0.195 0) in cell 9799 << Maximum is at exit after all heatingsurfaceFieldValue massFlow_in write:
sum("inlet") of phi = -1.999996 << Intake is 2 kg/s as plannedsurfaceFieldValue massFlow_out write:
sum("outlet") of phi = 1.773396 << Exit flow is below intake so mass is accumulating in pipesurfaceFieldValue p_inlet_avg write:
areaAverage("inlet") of p = 735063.4 << This is the new measurement.<< My interpretation: There is more pressure at the intake due to intake feature forcing gas in
End
ChatGPT5.2 has a recommendation for more information gathering.
inletForces
{
type forces;
libs ("libforces.so");// Only measure force on the inlet patch:
patches (inlet);// Field names (match your case)
pName p;
UName U;// Compressible: tell it which density field to use
rhoName rho;
rhoInf 1; // reference (won’t matter much here)// Optional: set a reference point (moments use it; forces don’t care)
CofR (0 0 0);// Output control:
writeControl writeTime;
writeInterval 1;
log true;
}
(th)
Offline
Like button can go here
#277 2026-01-21 15:29:28
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
A run from 4 seconds to 8 seconds just ended. This run included additional data gathering functions.
I don't know how to interpret the results but hope there is something useful here:
ExecutionTime = 11246.17 s ClockTime = 11269 s
volFieldValue p_avg write:
volAverage(heaterCells) of p = 665639.9volFieldValue T_avg write:
volAverage(heaterCells) of T = 757.2487volFieldValue rho_mass write:
volIntegrate(heaterCells) of rho = 4.465979volFieldValue T_min_global write:
min() of T = 26.83986 at location (-239.5 0.175 2.292485e-17) in cell 8880volFieldValue T_max_global write:
max() of T = 1789.563 at location (-0.5 -0.195 -2.162287e-17) in cell 239surfaceFieldValue massFlow_in write:
sum("inlet") of phi = -2surfaceFieldValue massFlow_out write:
sum("outlet") of phi = 1.893913surfaceFieldValue p_inlet_avg write:
areaAverage("inlet") of p = 676283forces inletForces write:
sum of forces:
pressure : (-27051.32 0 0)
viscous : (1.319819e-06 9.629071e-18 9.153922e-30)
porous : (0 0 0)
sum of moments:
pressure : (0 -4.582672e-15 1.051603e-11)
viscous : (3.828871e-24 2.240422e-25 -2.313687e-15)
porous : (0 0 0)End
(th)
Offline
Like button can go here
#278 2026-01-21 15:34:07
- SpaceNut
- Administrator
- From: New Hampshire
- Registered: 2004-07-22
- Posts: 30,284
Re: OpenFOAM
I was hoping that this had been read already but here it is About OpenFOAM
A solar thermal optical plane space vessel uses focused sunlight via large mirrors/concentrators, often with optical fibers, to heat a propellant (like hydrogen) to extreme temperatures, creating thrust, offering high efficiency (high specific impulse) for orbit changes, unlike solar sails that use light pressure, with concepts like Portal Space Systems' Supernova aiming for high maneuverability for defense/service, building on NASA's research for advanced solar propulsion systems.
Core Concepts
Solar Thermal Propulsion (STP): Captures solar energy with large mirrors (concentrators) and focuses it onto a heat exchanger/absorber, heating a propellant to generate thrust, achieving higher efficiency than chemical rockets.
Optical Waveguides: Some designs use optical fibers to transmit concentrated solar energy from the collector to the engine, allowing for more flexible system designs.
Optical Plane: Refers to the precise optical systems (mirrors, lenses) used to concentrate sunlight and potentially for imaging/sensing, requiring high-temperature materials and stability.
Key Components & Technologies
Concentrators: Large, often inflatable mirrors that focus sunlight.
Absorber/Heat Exchanger: Receives concentrated solar energy and heats the propellant.
Propellant: Typically hydrogen, heated to over 2500K, sometimes over 3000K with advanced materials.
Materials: Requires high-temperature refractory metals (tungsten, rhenium) or carbides to withstand extreme heat.
Optical Structures: Precision-built composite structures for stability in imaging and concentrating systems, used in telescopes and rovers.
Applications & Examples
Upper Stages: Ideal for moving payloads from Low Earth Orbit (LEO) to higher orbits or Earth escape.
Small Satellites: Enables orbit changes and long-duration missions with high delta-V.
Portal Supernova: A current project using solar thermal propulsion for high-delta-V, in-orbit refueling, and autonomous drone swarms.
NASA Research: Historically studied by NASA Marshall Space Flight Center (MSFC) and Sandia National Labs for advanced propulsion.
Contrast with Solar Sails
Solar Sails: Use direct photon pressure from sunlight on large sails, no propellant needed (e.g., IKAROS, LightSail-2).
Solar Thermal: Heats a propellant for rocket thrust; distinct from solar sails but both use solar energy for propulsion
Offline
Like button can go here
#279 2026-01-21 18:55:24
- SpaceNut
- Administrator
- From: New Hampshire
- Registered: 2004-07-22
- Posts: 30,284
Re: OpenFOAM
Solar thermal propulsion (STP) utilizing a hydrogen propellant, enhanced by heat pipe technology and leveraging high-performance, regeneratively cooled engine techniques (similar to the Merlin engine's regenerative cooling), is a highly efficient propulsion concept for in-space transportation. This approach, often characterized as a Solar Powered Rocket with Impulsive Thermal Engine (SPRITE), can achieve specific impulse (\(I_{sp}\)) values of 830–1000+ seconds, roughly twice that of the best chemical engines. Key Components &
Technology Hydrogen Propellant: Used for its low molecular weight, providing superior \(I_{sp}\) (800-1000+ seconds) compared to other propellants.
Heat Pipe/Thermal Storage: High-temperature sodium wicking or phase-change materials (PCM) are used to store solar energy, enabling the system to "coast" and accumulate energy, then release it for high-thrust, intermittent burns.
Materials: Refractory metals such as tungsten or rhenium are required to handle high temperatures (>2500K) and prevent chemical degradation from hot hydrogen.
"Merlin" Capabilities (Regenerative Cooling): Similar to the SpaceX Merlin engine, which uses regenerative cooling to handle extreme temperatures, solar thermal engines use the cold hydrogen propellant to cool the engine chamber and nozzle before it is heated and expelled.
Performance Characteristics Thrust & Efficiency: While early designs produced around 8.9 N of thrust, advancements in high-temperature materials and concentrator technology are aimed at increasing performance.
Energy Collection: Uses deployable solar concentrators to gather solar energy over time, eliminating the need to have collectors pointed at the sun during all maneuvers.
Propellant Management: To handle cryogenic hydrogen, advanced zero-boil-off (ZBO) techniques are used to ensure long-term storability.
Operational Advantages High Specific Impulse (\(I_{sp}\)): Over 800s to 1000s, significantly higher than chemical (\(I_{sp}\) ~300-450s).
No Hazardous Materials: Unlike nuclear thermal propulsion (NTP), solar thermal systems are generally non-toxic and safer to operate.High-
Thrust Capability: The use of energy storage/impulsive thermal engines allows for high-thrust maneuvers suitable for orbital transfers (e.g., LEO to GEO in ~1 month).
Versatility: The system is ideal for in-space tugs, utilizing on-orbit refueling or potentially harvested water/hydrogen as propellant. Current research, such as the NASA-supported SPRITE project, is focused on reducing the total wet mass of these vehicles to below 200 kg while maximizing propellant efficiency
Solar thermal propulsion (STP) using hydrogen heat pipes represents a high-efficiency alternative to traditional chemical engines like SpaceX's Merlin. While STP systems excel in fuel efficiency (Isp), they operate at vastly different scales than the high-thrust Merlin engines used for heavy lifting.
1. Comparative Capabilities (STP vs. Merlin)
As of 2026, experimental STP systems are achieving performance metrics that surpass the efficiency of Merlin engines while trailing significantly in raw power.Specific Impulse (Efficiency): STP systems such as the SPRITE (Solar Powered Rocket with Impulsive Thermal Engine) achieve an Isp of ~830–860 seconds. Advanced designs using carbide materials aim for >1,000 seconds. In contrast, the Merlin 1D vacuum engine has an Isp of ~348 seconds.
Thrust-to-Weight Ratio: The Merlin 1D holds a world record thrust-to-weight ratio of >150. While STP systems like SPRITE are designed for high thrust-to-weight relative to other solar concepts, they generally produce low absolute thrust (e.g., 8.9 N) suitable for orbital transfers rather than Earth-to-orbit launches.
Mission Utility: STP is primarily targeted for upper-stage maneuvers, such as moving payloads from LEO to GEO or the Moon, where high efficiency over long durations is more valuable than the immediate raw thrust required by first-stage engines like the Merlin.2. Role of Hydrogen Heat Pipes in 2026
In 2026, hydrogen heat pipes are critical for thermal management in these high-temperature systems:Thermal Transport: Heat pipes function as "thermal superconductors," using phase-change fluids to move massive heat loads from solar concentrators to the engine's absorber cavity with minimal temperature drops.
System Enhancement: Recent developments show that integrating parabolic reflectors with advanced heat pipes can increase local temperatures by over 30%, significantly improving the melting performance of phase-change materials used for energy storage.
Propellant Management: Solar thermal rockets utilize hot hydrogen as a propellant because its low molecular weight allows for much higher exhaust velocities than the kerosene/oxygen mix used in Merlin engines.
3. Current Milestones (2025–2026)
Full-Scale Tests: In late 2025, companies like Portal Space Systems validated STP architectures at temperatures of 1,500°C (2,700°F) using electrical induction to simulate solar power, with full-scale demonstration missions scheduled for 2026.
Advanced Materials: Research in 2026 continues to focus on refractory metals (tungsten, rhenium) and high-temperature carbides to withstand the 2,500K+ environments required for efficient hydrogen heating
Offline
Like button can go here
#280 2026-01-21 19:12:28
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
For all ... this topic is intended to serve as a tutorial for persons who are studying OpenFOAM for school, work or for personal projects.
We have attempted to report our experiences with OpenFOAM with as much detail as possible.
We have reported some successes, along with many failures.
A person should be able to read the topic from the top, and learn about various kinds of situations where OpenFOAM works well and others where it does not.
As of today, Wednesday 2026/01/21, we have a well debugged model that shows the original Merlin engine running with a 240 meter hydrogen heater in place of the original Merlin combustion chamber. We are working on a modification with a smaller nozzle throat, better suited for use with hydrogen as the propellan.
At the moment, we are debugging a case that is reporting ridiculous levels of thrust. The most likely cause for this misbehavior is a feature of OpenFOAM that generates artificial energy flow without the experimenter being aware of it. The feature is no doubt useful, but it is not helpful in every case.
For a practical engine, we will need to use some of the solar power (40 MW in the present case) to drive the pumps, so whatever is used to drive the pumps is not available to heat the gas.
(th)
Offline
Like button can go here
#281 Yesterday 07:16:20
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
It is time to begin thinking about adding a nozzle.
Up to now, our model consists of cubes 10 centimeters on a side. We have 9600 cubes making up the heating pipe, and 200 making up the stub. ChatGPT5.2 reminded me that when we introduce curves we will need to greatly reduce the X interval. At the moment I am thinking of millimeter slices for curves, and centimeter slices for straight sections.
In this part of the model, I see two curves at a minimum. The first is the in-bending curve to make the transition from the heating tubes to the straight wall that leads to the throat. The throat is one continuous curve that begins at the slope from the heating tubes, and ends after an approximately 120 degree bend to join a straight section for the expansion wedge.
Guesses: 30 degrees bend to straight, 120 degrees at throat, 30 degrees straight for expansion wedge
Please note the terminology. We will not have an expansion bell, because we have a 40 centimeter wide heating tube feeding the nozzle. Rather than try to collect all the gas from the heater into a cylinder, we have chosen to keep the 40 centimeter width of the heater through the nozzle.
The gap at the throat will be something like 7 millimeters. The idea is to achieve the area for the throat equivalent to the area of a circle of 3 cm diameter.
GW Johnson has provided a drawing to show the design he would like to see. This location is available if we can obtain a copy of the image.
Please note that in the version I am planning for the NewNozzle case, I will attempt to provide a circular merge between the heat pipe wall and the 30 degree slope down to the throat. While GW indicates we don't have to worry about that, i do not want to see the turbulence in the OpenFOAM animation that square corners would cause. The consume machine cycles and thus extend run time.
(th)
Offline
Like button can go here
#282 Yesterday 10:41:11
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
This post is about the latest run of NewNozzle from 8 seconds to 9 seconds.
ChatGPT5 is hot on the trail of our mysterious bug. It estimated the power needed to feed propellant into the heater at about 150 KW. I can understand that estimate, because the area of the pipe needed to feed hydrogen into the intake is small compared to that huge 40 centimeter by 10 centimeter back wall. All the thrust produced by the heater is pushing against that forward wall, so the propellant pumps have to push against that to inject gas into the chamber. However, since the area of the feed pipe is small compared to the total area of the forward wall, the pump will only be pushing against a fraction of the pressure on that forward wall. Since 150 KW is so much less than 40 MW I think we can ignore it for now.
here is one of the new functions we are adding to the case:
p_outlet_avg
{
type surfaceFieldValue;
libs ("libfieldFunctionObjects.so");
regionType patch;
name outlet;
operation areaAverage;
fields (p);
writeControl timeStep;
writeInterval 1;
log true;
}(th)
Offline
Like button can go here
#283 Yesterday 12:37:36
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
We have results from the run from 8 to 9 seconds.
I don't know what the results mean but someone else may.
ExecutionTime = 2844.41 s ClockTime = 2847 s
volFieldValue p_avg write:
volAverage(heaterCells) of p = 669188.7volFieldValue T_avg write:
volAverage(heaterCells) of T = 738.8852volFieldValue rho_mass write:
volIntegrate(heaterCells) of rho = 4.551232volFieldValue T_min_global write:
min() of T = 26.8307 at location (-239.5 0.175 2.292485e-17) in cell 8880volFieldValue T_max_global write:
max() of T = 1741.948 at location (-0.5 -0.195 -2.162287e-17) in cell 239surfaceFieldValue massFlow_in write:
sum("inlet") of phi = -2surfaceFieldValue massFlow_out write:
sum("outlet") of phi = 1.945187surfaceFieldValue outlet_U_avg write:
areaAverage("outlet") of U = (443.816 -1.602165e-08 1.816319e-15)surfaceFieldValue p_outlet_avg write:
areaAverage("outlet") of p = 658078.8surfaceFieldValue p_inlet_avg write:
areaAverage("inlet") of p = 679982.6forces inletForces write:
sum of forces:
pressure : (-27199.3 0 0)
viscous : (1.313604e-06 3.83113e-18 9.104782e-30)
porous : (0 0 0)
sum of moments:
pressure : (0 -4.607741e-15 5.712764e-12)
viscous : (3.803829e-24 2.229902e-25 -9.209213e-16)
porous : (0 0 0)End
ChatGPT5.2 and i are on a quest to try to understand the invalid results we are seeing from a previously (apparently) reliable program "thrustCalc".
We moved a lot of the extendedMerlin case over to the NewNozzle case, and there may be further adjustments needed. Right now, thrustCalc is showing ridiculous numbers for thrust. I understand that physics equations predict how much thrust is possible given 2 kg/s of hydrogen and 40 MW of energy.
We have seen figures that match the predictions when we ran extendedMerlin. However, with NewNozzle we are seeing ridiculous numbers.
What is curious about this situation is that we do NOT have a nozzle installed yet. This was supposed to be a shake-down run with nothing but a heat pipe and an empty stub where the nozzle should be. I am looking forward to finding out what-the-heck happened.
Here is the plot generated from the data provided by the run of NewNozzle from 0 to 9 seconds. The plot does not reflect reality. On the other hand, the plot itself accurately reflects the numbers it was given.
(th)
Offline
Like button can go here
#284 Yesterday 21:00:57
- tahanson43206
- Moderator
- Registered: 2018-04-27
- Posts: 23,960
Re: OpenFOAM
This is a detective story .... ChatGPT5.2 and I are following the clues to try to find the cause of the ridiculous results shown in the thrust plot.
We ** just ** learned that we had created the program some time ago. We both forgot about it. ChatGPT5 forgot because it is limited in the amount of memory it's managers allow it to retain. I forgot because I'm human.
Here is the likely cause of the incorrect results!
cat thrustCalc.C
// thrustCalc.C — OpenFOAM v12-compatible, standalone thrust post-processor
// Computes force on the FLUID across a patch:
// T = ∫_patch [ rho*(U·nHat)*U dA + (p - pAmb)*Sf ]
// Vehicle thrust = negative of this vector.
#include "argList.H"
#include "timeSelector.H"
#include "Time.H"
#include "fvMesh.H"
#include "volFields.H"
#include "surfaceFields.H"
#include "fvc.H"
#include "IOdictionary.H"
#include "polyBoundaryMesh.H"
#include "word.H"
#include "vector.H"
#include "OFstream.H"
#include "OSspecific.H" // mkDir
#include <fstream>
#include <iomanip>
using namespace Foam;
int main(int argc, char *argv[])
{
// Enable time-selection flags before args are created by setRootCase.H
timeSelector::addOptions();
// Nice help text with --help
argList::addNote
(
"Compute thrust on a patch:\n"
" T = ∫_patch [ rho*(U·nHat)*U dA + (p - pAmb)*Sf ]\n"
"Force reported is on the FLUID; vehicle thrust is its negative."
);
#include "setRootCase.H"
#include "createTime.H"
instantList timeDirs = timeSelector::select0(runTime, args);
#include "createMesh.H"
// Read settings from system/thrustCalcDict
IOdictionary dict
(
IOobject
(
"thrustCalcDict",
runTime.system(),
runTime,
IOobject::MUST_READ,
IOobject::NO_WRITE
)
);
word patchName("outlet");
dict.readIfPresent("patch", patchName);
word inletName(""); // leave absent to ignore
dict.readIfPresent("inletPatch", inletName);
scalar pAmb(0.0);
dict.readIfPresent("pAmb", pAmb);
// Output path
fileName outDir = runTime.path()/"postProcessing"/"thrustCalc"/patchName;
mkDir(outDir);
const fileName outFile = outDir/"thrust.dat";
// Helper: find patch index by name (v12-safe)
auto findPatch = [&](const word& name) -> label
{
if (!name.size()) return -1;
const fvBoundaryMesh& bnd = mesh.boundary();
forAll(bnd, i)
{
if (bnd[i].name() == name) return i;
}
return -1;
};
forAll(timeDirs, ti)
{
runTime.setTime(timeDirs[ti], ti);
mesh.readUpdate();
// Exact time directory name (no rounding)
const word curTName = timeDirs[ti].name();
// Read fields at this time
volVectorField U
(
IOobject("U", curTName, mesh, IOobject::MUST_READ, IOobject::NO_WRITE),
mesh
);
volScalarField p
(
IOobject("p", curTName, mesh, IOobject::MUST_READ, IOobject::NO_WRITE),
mesh
);
volScalarField rho
(
IOobject("rho", curTName, mesh, IOobject::MUST_READ, IOobject::NO_WRITE),
mesh
);
// Compute contribution on a patch (fields are in scope here)
auto computePatch = [&](const word& name, bool subtract) -> vector
{
if (!name.size()) return vector::zero; // silent ignore
vector result(vector::zero);
const label pid = findPatch(name);
if (pid < 0)
{
Info<< "thrustCalc: patch '"<< name <<"' not found (skipping)." << nl;
return result;
}
const vectorField& Sf = mesh.Sf().boundaryField()[pid];
const fvPatchVectorField& Ub = U.boundaryField()[pid];
const fvPatchScalarField& pb = p.boundaryField()[pid];
const fvPatchScalarField& rb = rho.boundaryField()[pid];
const label N = Sf.size();
for (label i=0; i<N; ++i)
{
const scalar magSf = mag(Sf[i]);
const vector nHat = (magSf > VSMALL ? Sf[i]/magSf : vector::zero);
const scalar Un = (Ub[i] & nHat);
const vector mom = rb[i]*Un*Ub[i]*magSf; // rho * Un * U * dA
const vector pTerm = (pb[i] - pAmb)*Sf[i]; // (p - pAmb) * Sf
result += (mom + pTerm);
}
return subtract ? -result : result;
};
const vector thrust = computePatch(patchName, /*subtract*/false)
+ computePatch(inletName, /*subtract*/true);
const scalar t = runTime.value();
const scalar Fmag = mag(thrust);
// Decide if we need a header (file does not exist or empty)
bool needHeader = true;
{
std::ifstream chk(outFile.c_str());
needHeader = !chk.good() || (chk.peek() == std::ifstream::traits_type::eof());
}
// Append one row (with header if needed)
std::ofstream os(outFile.c_str(), std::ios_base::app);
if (needHeader)
{
os << "# timeDir t[s] Fx[N] Fy[N] Fz[N] |F|[N]\n";
}
os.setf(std::ios_base::scientific);
os << curTName << " "
<< std::setprecision(10) << t << " "
<< thrust.x() << " " << thrust.y() << " " << thrust.z() << " "
<< Fmag << "\n";
os.close();
Info<< "thrustCalc: t="<< t
<< " Thrust_on_fluid [N] = " << thrust
<< " |F|=" << Fmag
<< " (vehicle reaction = -value above)" << nl;
}
Info<< "thrustCalc: wrote " << outFile << nl;
return 0;
}(th)
Offline
Like button can go here