Improve ElastoGap

Closes #4197

Prefer code to pre, and move things.

Add scaling ratio variable and add parameter description

Simplify computation of f_d

Documentation: add non-braking spaces where needed

Add scaling ratio variable and add parameter description

Simplify computation of f_d

Documentation: add non-braking spaces where needed

Improve documentation concerning s_rel and f_rel

Enable only positive values of s_rel and f_rel

Author: HansOlsson

Modelica/Mechanics/Translational/Components/ElastoGap.mo

@@ -6,8 +6,10 @@ model ElastoGap "1D translational spring damper combination with gap"
   parameter SI.TranslationalDampingConstant d(final min=0, start=1)
     "Damping constant";
   parameter SI.Position s_rel0=0 "Unstretched spring length";
+  parameter SI.Force f_ref(min=0) = c*s_ref "Reference spring force at s_ref" annotation(Dialog(tab="Advanced"));
+  parameter SI.Length s_ref(min=Modelica.Constants.eps) = 1 "Reference relative compression at which f_c = f_ref" annotation(Dialog(tab="Advanced"));
   parameter Real n(final min=1) = 1
-    "Exponent of spring force ( f_c = -c*|s_rel-s_rel0|^n )";
+    "Exponent of spring force ( f_c = -f_ref*|(s_rel-s_rel0)/s_ref|^n )";
   extends Modelica.Thermal.HeatTransfer.Interfaces.PartialElementaryConditionalHeatPortWithoutT;
 
   /*
@@ -22,14 +24,15 @@ protected
   SI.Force f_d2 "Linear damping force";
   SI.Force f_d
     "Linear damping force which is limited by spring force (|f_d| <= |f_c|)";
+  Real ratio "Scaling ratio of relative compression to s_ref";
 equation
   // Modify contact force, so that it is only "pushing" and not
   // "pulling/sticking" and that it is continuous
   contact = s_rel < s_rel0;
-  f_c = smooth(1, noEvent(if contact then -c*abs(s_rel - s_rel0)^n else 0));
+  ratio = (s_rel - s_rel0)/s_ref;
+  f_c = smooth(1, noEvent(if contact then -f_ref*abs(ratio)^n else 0));
   f_d2 = if contact then d*v_rel else 0;
-  f_d = smooth(0, noEvent(if contact then (if f_d2 < f_c then f_c else if
-    f_d2 > -f_c then -f_c else f_d2) else 0));
+  f_d = smooth(0, noEvent(if contact then min(max(f_d2, f_c), -f_c) else 0));
   f = f_c + f_d;
   lossPower = f_d*v_rel;
   annotation (
@@ -49,7 +52,7 @@ a nonlinear spring force can be modeled:
 </p>
 
 <blockquote><pre>
-desiredContactForce = c*|s_rel - s_rel0|^n + d*<strong>der</strong>(s_rel)
+desiredContactForce = f_ref*|(s_rel - s_rel0)/s_ref|^n + d*<strong>der</strong>(s_rel)
 </pre></blockquote>
 
 <p>
@@ -112,8 +115,23 @@ If s_rel0 = 0, the equations are:
 </pre></blockquote>
 
 <p>
-Note, since |f_d| &le; |f_c|, pulling forces cannot occur and the contact force
-is always continuous, especially around the start of the penetration at s_rel = s_rel0.
+Note, since |f_d|&nbsp;&le;&nbsp;|f_c|, pulling forces cannot occur and the contact force
+is always continuous, especially around the start of the penetration at s_rel&nbsp;= s_rel0.
+On the contrary, this leads to the contact force <code>f&nbsp;=&nbsp;0</code> even if
+<code>contact&nbsp;= true</code> is still indicated around the end of the penetration.
+This is because <code>contact</code> indicates only the occurance of geometry penetration.
+</p>
+
+<p>
+In order to have consistent units for nonlinear springs, the term <code>c*|s_rel|^n</code>
+is replaced by <code>f_ref*|s_rel/s_ref|^n</code>, whereby <code>s_ref</code> is a&nbsp;reference
+length for the spring and <code>f_ref</code> is the spring force when
+<code>s_rel&nbsp;=&nbsp;s_ref</code>. The default values <code>s_ref&nbsp;=&nbsp;1</code>
+and <code>f_ref&nbsp;= c*s_ref</code> lead to the same results of the two above-mentioned terms.
+Setting the advanced parameters <code>s_ref</code> and <code>f_ref</code> for a&nbsp;nonlinear
+spring directly gives a&nbsp;cleaner and straightforward parametrization.
+For simplicity reasons, both <code>s_ref</code> and <code>f_ref</code> are considered being
+positive.
 </p>
 
 <p>