Update range-solver-lib with contaminant dispersion fix and correct time-dependent value docs

Author: tomas-soltys

src/fea/doc/RELEASE_NOTES.md

@@ -218,6 +218,21 @@
   accepted alternatives. An acoustic entity carrying a density and a speed of
   sound is no longer flagged for missing a modulus of elasticity
 
+#### Contaminant dispersion
+
+- A transient contaminant dispersion run no longer blows up. The particle
+  concentration used to grow far above the highest inflow value, first where
+  the flow recirculates, because of errors in the stabilisation of the
+  equations and because negative values were clipped to zero after every
+  time step. On the tutorial channel model the maximum now follows the inflow
+  value (`0.618` at time-step 618, where `3326` was reported before).
+  Negative concentrations now appear instead of being hidden by the clipping;
+  they are numerical undershoots, not missing contaminant, and are largest at a
+  few nodes
+- The *Contaminant dispersion in fluids* tutorial states that time-dependent
+  values are interpolated linearly between the given times. It used to say
+  that each value applies from its time onwards
+
 #### Solver control
 
 - **Stop solver** and **Kill solver** are enabled only while a solver is

src/fea/help/theory/en/theory_fluid.md

@@ -961,7 +961,9 @@ domain, `L / |v|`, divided by that step.
    `0`, the release concentration at the moment the release starts, and `0`
    again when it ends.
 
-Each value is valid **from** the time given. This time-triggered condition is
+Between two given times the value is interpolated linearly, and after the last
+time the last value is kept, so a release that should start abruptly needs two
+entries close together. This time-dependent condition is
 what turns a steady inlet into a release event, and it is the heart of a
 dispersion model.
 
@@ -1006,7 +1008,7 @@ the march. Without it the run starts from rest and the first steps are wasted.
 |---|---|
 | The concentration never leaves the inlet | the flow field is zero - the flow task is missing from the group, or the run was not restarted from a converged flow |
 | The plume smears out almost immediately | the Courant number is far above 1, or the mesh is too coarse along the path |
-| The concentration goes negative or overshoots | the time step is too large for the mesh, so the stabilised advection is ringing |
+| The concentration goes negative or overshoots | the time step is too large for the mesh, so the stabilised advection is ringing. Slight undershoots next to a steep front are normal; the solution is not clipped at zero, because clipping would add contaminant |
 | Only one record was written | the output frequency is `0`, which writes the last step only |
 | Each step's residual stays high | too few iterations per time step - the flow within each step is not converged |
 | The plume never spreads sideways | that is the model: the diffusivity is zero, so lateral spreading comes only from the flow field |

src/fea/help/theory/sk/theory_fluid.md

@@ -961,7 +961,9 @@ domain, `L / |v|`, divided by that step.
    `0`, the release concentration at the moment the release starts, and `0`
    again when it ends.
 
-Each value is valid **from** the time given. This time-triggered condition is
+Between two given times the value is interpolated linearly, and after the last
+time the last value is kept, so a release that should start abruptly needs two
+entries close together. This time-dependent condition is
 what turns a steady inlet into a release event, and it is the heart of a
 dispersion model.
 
@@ -1006,7 +1008,7 @@ the march. Without it the run starts from rest and the first steps are wasted.
 |---|---|
 | The concentration never leaves the inlet | the flow field is zero - the flow task is missing from the group, or the run was not restarted from a converged flow |
 | The plume smears out almost immediately | the Courant number is far above 1, or the mesh is too coarse along the path |
-| The concentration goes negative or overshoots | the time step is too large for the mesh, so the stabilised advection is ringing |
+| The concentration goes negative or overshoots | the time step is too large for the mesh, so the stabilised advection is ringing. Slight undershoots next to a steep front are normal; the solution is not clipped at zero, because clipping would add contaminant |
 | Only one record was written | the output frequency is `0`, which writes the last step only |
 | Each step's residual stays high | too few iterations per time step - the flow within each step is not converged |
 | The plume never spreads sideways | that is the model: the diffusivity is zero, so lateral spreading comes only from the flow field |

src/fea/help/tutorials/en/tutorial_Contaminant_dispersion_in_fluids.md

@@ -126,7 +126,7 @@ Apply **Particle concentration** boundary condition to **Inflow** model entity.
 
 Do not specify value but expand the **Particle concentration** property (arrow in front of its name) and click on **Edit time dependent values** button, to specify time-triggered (time-profile) boundary condition.
 
-In the **Component editor** dialog, time-dependent values can be specified. Values are always valid **from** the specified time. Set **Number of values** to **3** and enter values as shown below (double click on a cell to edit it), then click **Ok**.
+In the **Component editor** dialog, time-dependent values can be specified. Between two specified times the value is interpolated linearly, and after the last specified time the last value is kept. With the values below the inflow concentration therefore rises from 0 to 1 `[kg/m^3]` during the first 10 seconds and falls back to 0 between 10 and 15 seconds. Set **Number of values** to **3** and enter values as shown below (double click on a cell to edit it), then click **Ok**.
 
 ![CFD - Particle concentration condition](image-CFD-Particle_concentration_condition.png)
 

src/fea/help/tutorials/sk/tutorial_Contaminant_dispersion_in_fluids.md

@@ -76,7 +76,7 @@ Aplikujte okrajovú podmienku **Koncentrácia častíc** na entitu modelu **Prí
 
 Nezadávajte hodnotu, ale kliknite na tlačidlo **Upraviť časovo závislé hodnoty** na zadanie časovo spustenej (časový profil) okrajovej podmienky.
 
-V dialógu **Editor komponentov** možno zadávať časovo závislé hodnoty. Hodnoty sú vždy platné **od** zadaného času.
+V dialógu **Editor komponentov** možno zadávať časovo závislé hodnoty. Medzi dvoma zadanými časmi sa hodnota lineárne interpoluje a po poslednom zadanom čase zostáva platná posledná hodnota. Pri hodnotách uvedených nižšie preto koncentrácia na vstupe počas prvých 10 sekúnd stúpa z 0 na 1 `[kg/m^3]` a medzi 10. a 15. sekundou klesne späť na 0.
 
 ![CFD – Podmienka koncentrácie častíc](image-CFD-Particle_concentration_condition.png)