---
id: model-06
title: "Part 4 Review and Comprehensive Modeling Project"
section: "Advanced Modeling"
difficulty: "advanced"
estimated_time: 90
prerequisites: ["model-01", "model-02", "model-03", "model-04", "model-05"]
objectives:
  - Synthesize all advanced modeling concepts from Part 4
  - Apply complete workflow from FEMM to validated spark model
  - Compare lumped vs distributed approaches systematically
  - Execute comprehensive modeling project integrating all skills
tags: ["review", "integration", "project", "validation", "comprehensive"]
---

# Part 4 Review and Comprehensive Modeling Project

This lesson reviews all advanced modeling concepts from Part 4 and guides you through a comprehensive project that integrates FEMM extraction, circuit implementation, resistance optimization, and validation.

## Part 4 Concepts Summary

### Lesson 1: Lumped Model Theory

**Key concepts:**
```
Structure: C_mut - R - C_sh network
  - C_mut: Topload to spark coupling
  - R: Effective plasma resistance
  - C_sh: Spark to ground shunt

When to use:
  ✓ Sparks <1-2 m
  ✓ Impedance matching studies
  ✓ Quick design iterations
  ✓ Engineering estimates

Workflow:
  1. FEMM electrostatic (2-body)
  2. Extract C_mut, C_sh from 2×2 matrix
  3. Calculate R = 1/(ω × C_total)
  4. Build SPICE, simulate
  5. Validate: φ_Z, R range, C_sh ≈ 2 pF/ft
```

### Lesson 2: FEMM Extraction - Lumped

**Key concepts:**
```
Maxwell matrix convention:
  - Diagonal: C_ii > 0 (self-capacitance)
  - Off-diagonal: C_ij < 0 (mutual, negative!)
  - Symmetric: C_ij = C_ji
  - Row sum ≈ 0 (ground at infinity)

Extraction formulas:
  C_mut = |C₁₂|           (absolute value!)
  C_sh = C₂₂ - |C₁₂|      (subtract absolute)

Sign convention critical:
  - Maxwell: negative off-diagonals
  - Circuit: positive capacitances
  - Conversion: Take absolute value

Validation:
  ✓ Symmetry <1% error
  ✓ C_sh ≈ 2 pF/ft ± factor 2
  ✓ Physical value ranges
  ✓ Ground distance sensitivity test
```

### Lesson 3: Distributed Model Theory

**Key concepts:**
```
Why distributed:
  - Long sparks (>2 m)
  - Current distribution matters
  - Leader/streamer transitions
  - Research applications

Segmentation:
  - Equal-length segments
  - n = 5-20 typical
  - Convergence test: double n

Circuit topology:
  - (n+1)×(n+1) capacitance matrix
  - n resistance values
  - O(n²) complexity

Physical expectations:
  - R monotonically increasing
  - Current decreasing base→tip
  - Voltage non-linear drop
  - Power concentrated at base

Trade-off: 1000-2000× slower than lumped
```

### Lesson 4: FEMM Extraction - Distributed

**Key concepts:**
```
Multi-body setup:
  - n conductors + topload
  - 0.1 mm gaps between segments
  - Consistent numbering critical

Matrix validation:
  ✓ Symmetry
  ✓ Positive semi-definite (passivity)
  ✓ Adjacent > distant coupling
  ✓ Total C_sh vs 2 pF/ft rule

SPICE implementation:
  1. Partial capacitance (flip signs)
  2. Controlled sources (direct)
  3. Nearest-neighbor (approximation)

C_sh discrepancy:
  - Factor 2-3 normal for distributed
  - Matrix method vs empirical rule
  - Use FEMM values (more accurate)
```

### Lesson 5: Resistance Optimization

**Key concepts:**
```
Iterative method:
  - Initialize: tapered profile
  - Optimize each R[i] sequentially
  - Apply damping (α ≈ 0.3-0.5)
  - Position-dependent bounds
  - Convergence: <1% change

Position-dependent bounds:
  R_min: 1 kΩ → 10 kΩ (base to tip)
  R_max: 100 kΩ → 100 MΩ (quadratic)

Simplified method:
  R[i] = 1/(ω × C_total[i])
  - 1000× faster
  - ±20% accuracy
  - Use for standard cases

Validation:
  ✓ R_total: 50-500 kΩ at 200 kHz
  ✓ Monotonic increase
  ✓ Scales as R ∝ 1/f, R ∝ L
```

## Complete Modeling Workflow Checklist

### Phase 1: Problem Definition

```
[ ] Define spark length L_total
[ ] Specify operating frequency f
[ ] Choose model type:
    [ ] Lumped (if L < 2 m)
    [ ] Distributed n=___ (if L ≥ 2 m)
[ ] Gather topload geometry data
[ ] Determine ground plane position
```

### Phase 2: FEMM Geometry and Solve

```
[ ] Create FEMM geometry:
    [ ] Axisymmetric (r-z)
    [ ] Topload (toroid/sphere)
    [ ] Spark segment(s)
    [ ] Ground plane
    [ ] Outer boundary
[ ] Define materials (Air, ε_r=1)
[ ] Assign conductors:
    [ ] Conductor 0: Topload, V=1V
    [ ] Conductors 1-n: Segments, floating
    [ ] Boundary: Ground, V=0
[ ] Generate mesh (check quality)
[ ] Solve electrostatic problem
[ ] Extract capacitance matrix [C]
```

### Phase 3: Matrix Validation

```
[ ] Check symmetry: |C[i,j] - C[j,i]| / |C[i,j]| < 0.01
[ ] Check diagonal positive: C[i,i] > 0 for all i
[ ] Check off-diagonal negative: C[i,j] < 0 for i≠j
[ ] Check passivity: Eigenvalues ≥ 0
[ ] Check physical patterns:
    [ ] Adjacent > distant coupling
    [ ] Topload coupling decreases with distance
[ ] Check total C_sh vs 2 pF/ft rule (factor 2-3 OK)
```

### Phase 4: Resistance Determination

```
[ ] Choose method:
    [ ] Iterative (research, extreme cases)
    [ ] Simplified (standard cases, engineering)

If Iterative:
[ ] Initialize tapered profile
[ ] Define position-dependent bounds
[ ] Set damping factor α
[ ] Run optimization loop
[ ] Check convergence (<1% or <5% for tip)
[ ] Validate R distribution (monotonic, ranges)

If Simplified:
[ ] Calculate C_total[i] for each segment
[ ] Compute R[i] = 1/(ω × C_total[i])
[ ] Apply bounds: R[i] = clip(R[i], R_min[i], R_max[i])
[ ] Validate total R_total (50-500 kΩ at 200 kHz)
```

### Phase 5: SPICE Implementation

```
[ ] Convert C matrix to SPICE format:
    [ ] Partial capacitances (most common)
    [ ] Or controlled sources (advanced)
    [ ] Or nearest-neighbor (approximation)
[ ] Add resistance elements R[i]
[ ] Define voltage source (test or from coil)
[ ] Set up AC analysis at operating frequency
[ ] Verify netlist syntax
```

### Phase 6: Simulation and Analysis

```
[ ] Run SPICE AC analysis
[ ] Extract results:
    [ ] Voltages V[i] at each node
    [ ] Currents I[i] through each segment
    [ ] Admittance Y_spark at topload
    [ ] Impedance Z_spark = 1/Y_spark
[ ] Calculate power distribution:
    [ ] P[i] = 0.5 × |I[i]|² × R[i]
    [ ] P_total = Σ P[i]
[ ] Plot distributions:
    [ ] V vs position
    [ ] I vs position
    [ ] P vs position
```

### Phase 7: Validation

```
[ ] Phase angle: -55° < φ_Z < -75°
[ ] Total resistance: 50-500 kΩ at 200 kHz
[ ] Current distribution: Decreasing base→tip
[ ] Voltage distribution: Non-linear, physical
[ ] Power balance: Concentrated at base
[ ] Compare to lumped model (if applicable)
[ ] Compare to measurements (if available)
```

### Phase 8: Documentation

```
[ ] Save FEMM geometry and results
[ ] Save capacitance matrix
[ ] Save resistance values
[ ] Save SPICE netlist
[ ] Save simulation results
[ ] Document validation checks
[ ] Record any issues/assumptions
```

## Lumped vs Distributed Comparison

### When Results Should Agree

**Equivalent impedance at topload:**
```
Lumped: Z_spark = R + 1/(jωC_total)
Distributed: Z_spark (from network)

Expected: Within 20-30% for well-designed models

Example:
  Lumped: |Z| = 180 kΩ ∠-70°
  Distributed: |Z| = 195 kΩ ∠-68°
  Difference: 8% ✓ Good agreement
```

**Total resistance:**
```
Lumped: Single R value
Distributed: R_total = Σ R[i]

Should be similar order of magnitude
Factor <2 difference: Excellent
Factor 2-3: Acceptable
Factor >5: Investigate
```

**Total capacitance:**
```
Lumped: C_total = C_mut + C_sh
Distributed: More complex (matrix network)

At topload, should see similar capacitive reactance
```

### When Results May Differ

**Current distribution:**
```
Lumped: Assumes uniform (no spatial info)
Distributed: Non-uniform, physically realistic

Cannot compare directly - distributed provides extra detail
```

**Power distribution:**
```
Lumped: Single power value (total)
Distributed: Spatial distribution P[i]

Lumped gives total only
Distributed shows WHERE power dissipated
```

**Tip behavior:**
```
Lumped: Averaged properties
Distributed: Can show tip streaming (low current, high R)

Distributed more realistic for long sparks
```

**Short spark (e.g., 0.8 m):**
```
Lumped and distributed should agree closely
Spatial variations small
Use lumped (simpler, faster)
```

**Long spark (e.g., 3 m):**
```
Distributed shows significant spatial variation
Lumped may over-predict tip current/power
Use distributed for accuracy
```

## Comprehensive Modeling Project

### Project Goal

**Design and model a complete spark system:**
```
Objective: Predict performance of 2.5 m spark at 200 kHz
Approach: Use distributed model (n=10)
Output: Current, voltage, power distributions + validation
```

### Project Specifications

```
Tesla coil system:
  - Operating frequency: f = 200 kHz
  - Topload: Toroid, 40 cm major dia, 12 cm minor dia
  - Target spark length: 2.5 m = 8.2 feet
  - Ground plane: 20 cm below spark tip
  - Topload voltage: 350 kV (estimate)

Model requirements:
  - Distributed model: n = 10 segments
  - Each segment: 0.25 m length
  - FEMM extraction: Full 11×11 matrix
  - Resistance: Simplified method
  - Validation: All checks
```

### Step 1: FEMM Setup

**Geometry parameters:**
```
Topload (toroid):
  - Major radius: 20 cm
  - Minor radius: 6 cm
  - Center at z = 0
  - Lowest point: z = -6 cm

10 spark segments:
  - Each length: 25 cm
  - Diameter: 2 mm (uniform)
  - Positions:
    Segment 1 (base): z = -6.1 to -31.1 cm
    Segment 2: z = -31.2 to -56.2 cm
    ...
    Segment 10 (tip): z = -231.5 to -256.5 cm

Ground plane:
  - z = -270 cm (20 cm below tip)
  - r = 0 to 400 cm

Outer boundary:
  - r = 400 cm
  - z = -300 to +50 cm
  - V = 0 boundary condition
```

**Expected mesh:**
```
Elements: 40,000-70,000
Refinement: 0.5 mm near spark, 50 mm at boundary
Solve time: 30-60 seconds
```

### Step 2: Matrix Extraction (Example Results)

**Hypothetical FEMM output (11×11 matrix):**

```
         [0]     [1]     [2]     [3]     [4]     [5]     [6]     [7]     [8]     [9]    [10]
[0]  [  38.2   -10.5    -4.2    -2.1    -1.2    -0.8    -0.5    -0.4    -0.3    -0.2    -0.1 ]
[1]  [ -10.5    16.2    -3.5    -1.4    -0.7    -0.4    -0.3    -0.2    -0.2    -0.1    -0.1 ]
[2]  [  -4.2    -3.5    12.8    -3.2    -1.3    -0.6    -0.4    -0.3    -0.2    -0.1    -0.1 ]
[3]  [  -2.1    -1.4    -3.2    11.4    -2.9    -1.2    -0.5    -0.3    -0.2    -0.1    -0.1 ]
[4]  [  -1.2    -0.7    -1.3    -2.9    10.6    -2.7    -1.1    -0.5    -0.3    -0.2    -0.1 ]
[5]  [  -0.8    -0.4    -0.6    -1.2    -2.7     9.8    -2.5    -1.0    -0.4    -0.2    -0.1 ]
[6]  [  -0.5    -0.3    -0.4    -0.5    -1.1    -2.5     9.2    -2.3    -0.9    -0.4    -0.1 ]
[7]  [  -0.4    -0.2    -0.3    -0.3    -0.5    -1.0    -2.3     8.6    -2.1    -0.8    -0.2 ]
[8]  [  -0.3    -0.2    -0.2    -0.2    -0.3    -0.4    -0.9    -2.1     8.2    -1.9    -0.5 ]
[9]  [  -0.2    -0.1    -0.1    -0.1    -0.2    -0.2    -0.4    -0.8    -1.9     7.6    -1.6 ]
[10] [  -0.1    -0.1    -0.1    -0.1    -0.1    -0.1    -0.1    -0.2    -0.5    -1.6     6.8 ] pF
```

*Note: These are illustrative values for the exercise.*

### Step 3: Matrix Validation

**Check symmetry:**
```
Example: C[2,5] = -0.6 pF, C[5,2] = -0.6 pF
Error: |(-0.6) - (-0.6)| / 0.6 = 0%
✓ Symmetric (check all pairs)
```

**Check patterns:**
```
Topload coupling: |C[0,1]| = 10.5 > |C[0,5]| = 0.8 > |C[0,10]| = 0.1 ✓
Adjacent coupling: |C[3,4]| = 2.9 > |C[3,7]| = 0.3 ✓
Diagonal positive: All C[i,i] > 0 ✓
Off-diagonal negative: All C[i,j] < 0 for i≠j ✓
```

**Total shunt capacitance:**
```
C_sh_total = Σᵢ₌₁¹⁰ (C[i,i] - |C[i,0]|)
           = (16.2-10.5) + (12.8-4.2) + ... + (6.8-0.1)
           = 5.7 + 8.6 + 9.3 + 9.4 + 9.8 + 9.7 + 9.6 + 9.4 + 9.3 + 6.7
           = 87.5 pF

Expected: 2 pF/ft × 8.2 ft = 16.4 pF

Ratio: 87.5 / 16.4 = 5.3

Higher than lumped expectation, but within factor 2-6 for distributed
Matrix method includes all couplings - acceptable ✓
```

### Step 4: Calculate Resistances (Simplified Method)

**Frequency:**
```
f = 200 kHz
ω = 2π × 200×10³ = 1.257×10⁶ rad/s
```

**Segment 1 (base):**
```
C_total[1] = |C[1,0]| + |C[1,2]| + ... + |C[1,10]|
           = 10.5 + 3.5 + 1.4 + 0.7 + 0.4 + 0.3 + 0.2 + 0.2 + 0.1 + 0.1
           = 17.4 pF

R[1] = 1 / (ω × C_total[1])
     = 1 / (1.257×10⁶ × 17.4×10⁻¹²)
     = 45.7 kΩ

Bounds: R_min[1] = 1 kΩ, R_max[1] = 100 kΩ
Check: 1 < 45.7 < 100 ✓
```

**Calculate similarly for all segments:**

```
Results (example):
R[1] = 45.7 kΩ   (position 0.00)
R[2] = 58.3 kΩ   (position 0.11)
R[3] = 71.2 kΩ   (position 0.22)
R[4] = 86.5 kΩ   (position 0.33)
R[5] = 105 kΩ    (position 0.44)
R[6] = 128 kΩ    (position 0.56)
R[7] = 157 kΩ    (position 0.67)
R[8] = 195 kΩ    (position 0.78)
R[9] = 248 kΩ    (position 0.89)
R[10] = 320 kΩ   (position 1.00)

Total: R_total = 1415 kΩ = 1.42 MΩ
```

**Validation:**
```
✓ Monotonically increasing
✓ Each within position-dependent bounds
✓ Total: Expected 50-500 kΩ, got 1.42 MΩ

Higher than typical - long spark (2.5 m), tip-dominated
Within factor 3-5 of estimates - acceptable for distributed model
```

### Step 5: Build SPICE Netlist

**Partial capacitance conversion (selected):**
```spice
* 10-segment distributed spark model - 2.5 m at 200 kHz
.param freq=200k

* Test voltage source
V_test topload 0 AC 1V

* Partial capacitances - between nodes (sample)
C_0_1 topload seg1 10.5p
C_0_2 topload seg2 4.2p
C_1_2 seg1 seg2 3.5p
C_2_3 seg2 seg3 3.2p
C_3_4 seg3 seg4 2.9p
* ... (continue for all pairs) ...

* Partial capacitances - to ground (sample)
C_0_gnd topload 0 {38.2 - (10.5+4.2+2.1+1.2+0.8+0.5+0.4+0.3+0.2+0.1)}
C_1_gnd seg1 0 {16.2 - (10.5+3.5+1.4+0.7+0.4+0.3+0.2+0.2+0.1+0.1)}
* ... (continue for all nodes) ...

* Resistances
R1 seg1 seg1_r 45.7k
R2 seg2 seg2_r 58.3k
R3 seg3 seg3_r 71.2k
R4 seg4 seg4_r 86.5k
R5 seg5 seg5_r 105k
R6 seg6 seg6_r 128k
R7 seg7 seg7_r 157k
R8 seg8 seg8_r 195k
R9 seg9 seg9_r 248k
R10 seg10 seg10_r 320k

* AC analysis
.ac lin 1 200k 200k

* Output
.print ac v(topload) v(seg1) v(seg2) v(seg3) v(seg4) v(seg5)
+         v(seg6) v(seg7) v(seg8) v(seg9) v(seg10)
.print ac i(V_test) i(R1) i(R2) i(R3) i(R4) i(R5)
+         i(R6) i(R7) i(R8) i(R9) i(R10)

.end
```

### Step 6: Simulation Results (Example)

**Voltage distribution (normalized, V_topload = 1V test):**
```
V[topload] = 1.000 V
V[seg1] = 0.842 V   (16% drop from topload)
V[seg2] = 0.714 V
V[seg3] = 0.608 V
V[seg4] = 0.518 V
V[seg5] = 0.441 V   (56% of topload)
V[seg6] = 0.375 V
V[seg7] = 0.318 V
V[seg8] = 0.269 V
V[seg9] = 0.227 V
V[seg10] = 0.192 V  (tip, 19% of topload)

Non-linear drop ✓ Expected for distributed capacitance
```

**Current distribution:**
```
I[seg1] = 18.4 μA   (base, highest)
I[seg2] = 12.2 μA   (66% of base)
I[seg3] = 8.54 μA
I[seg4] = 6.00 μA
I[seg5] = 4.20 μA   (23% of base)
I[seg6] = 2.93 μA
I[seg7] = 2.03 μA
I[seg8] = 1.38 μA
I[seg9] = 0.91 μA
I[seg10] = 0.60 μA  (tip, 3% of base)

Monotonically decreasing ✓ Capacitive shunting effect
```

**Power distribution:**
```
P[1] = 0.5 × (18.4×10⁻⁶)² × 45.7×10³ = 7.74 μW
P[2] = 0.5 × (12.2×10⁻⁶)² × 58.3×10³ = 4.34 μW
P[3] = 0.5 × (8.54×10⁻⁶)² × 71.2×10³ = 2.60 μW
...
P[10] = 0.5 × (0.60×10⁻⁶)² × 320×10³ = 0.058 μW

Total: P_total ≈ 21.5 μW (at 1V test)

Base segments (1-3): 14.7 μW (68% of total)
Middle (4-7): 5.8 μW (27%)
Tip (8-10): 1.0 μW (5%)

Power concentrated at base ✓ Physical expectation
```

**Impedance at topload:**
```
Y = I_test / V_test = 18.4 μA / 1V = 18.4 μS
|Z| = 1/18.4×10⁻⁶ = 54.3 kΩ
φ_Z ≈ -62° (calculated from Re{Y}, Im{Y})

Check: -55° < -62° < -75° ✓ Expected range
```

### Step 7: Scale to Actual Voltage

**Given: V_topload = 350 kV actual**

**Power scaling:**
```
P_actual = P_test × (V_actual / V_test)²
         = 21.5 μW × (350×10³ / 1)²
         = 21.5×10⁻⁶ × 1.225×10¹¹
         = 2.63 MW

Total power to spark: 2.63 MW
```

**Segment powers:**
```
P[1] = 7.74 μW × scale = 949 kW (36%)
P[2] = 4.34 μW × scale = 532 kW (20%)
P[3] = 2.60 μW × scale = 319 kW (12%)
...

Base heavily loaded, tip lightly loaded ✓
```

### Step 8: Final Validation

```
✓ Phase angle: φ_Z = -62° in range (-55° to -75°)
✓ Total resistance: 1.42 MΩ (high end, but acceptable for 2.5 m)
✓ Voltage distribution: Non-linear, physically reasonable
✓ Current distribution: Decreasing base→tip monotonically
✓ Power distribution: 68% in base 1/3, physical
✓ Matrix validation: All checks passed
✓ Resistance monotonic: Increasing base→tip

Model complete and validated!
```

## Key Takeaways from Part 4

- **Lumped models:** Fast (<1s), accurate for short sparks (<2 m), C_mut-R-C_sh structure
- **FEMM extraction:** Maxwell matrix has negative off-diagonals, C_mut = |C₁₂|, C_sh = C₂₂ - |C₁₂|
- **Distributed models:** Necessary for long sparks (>2 m), captures spatial variations, 1000× slower
- **Segmentation:** Equal lengths, n = 5-20, convergence test by doubling n
- **Matrix validation:** Symmetry, passivity (eigenvalues ≥ 0), physical patterns critical
- **SPICE implementation:** Partial capacitance method (flip signs), controlled sources, or nearest-neighbor
- **Resistance optimization:** Iterative (rigorous, slow) or simplified R = 1/(ωC) (fast, ±20%)
- **Position-dependent bounds:** R_min 1k→10k, R_max 100k→100M, prevents unphysical solutions
- **Validation ranges:** R_total 50-500 kΩ at 200 kHz typical, factor 2-3 variation acceptable
- **C_sh discrepancy:** Factor 2-3 from 2 pF/ft rule normal for distributed (use FEMM values)
- **Current distribution:** Decreases base→tip due to capacitive shunting (can be 20:1 ratio)
- **Power concentration:** 60-70% in base 1/3 of spark, tip contributes <10%

## Practice

{exercise:model-ex-06}

---
**Congratulations!** You have completed Part 4: Advanced Modeling. You now have the skills to:
- Build lumped spark models for quick analysis
- Extract capacitance matrices from FEMM for single and multi-body problems
- Construct distributed models for long sparks and research applications
- Optimize resistance distributions using iterative or simplified methods
- Validate models against physical expectations and measurements
- Apply complete modeling workflow from geometry to validated predictions

**Next Steps:**
- Part 5: Integration and Calibration (coming soon)
- Apply these techniques to your own Tesla coil designs
- Validate against measurements and refine models
- Contribute to the community knowledge base