diff --git a/src/core/placement.py b/src/core/placement.py
index bc9b0d5..7d84850 100644
--- a/src/core/placement.py
+++ b/src/core/placement.py
@@ -35,7 +35,22 @@ def get_coverage_weight(expected_demand: float) -> float:
 def covered_cells(h3_cell: str, radius: int) -> Set[str]:
     """Obtiene todas las celdas H3 cubiertas por un radio dado."""
     try:
-        return set(h3.k_ring(h3_cell, radius))
+        # H3 >= 4.x (API strings)
+        try:
+            import h3.api.basic_str as h3_api
+            return set(h3_api.grid_disk(h3_cell, radius))
+        except ImportError:
+            pass
+
+        # H3 3.x
+        try:
+            import h3
+            return set(h3.k_ring(h3_cell, radius))
+        except Exception:
+            pass
+
+        raise RuntimeError("API H3 no compatible")
+
     except Exception as e:
         print(f"Error calculando cobertura para celda {h3_cell}: {e}")
         return {h3_cell}
@@ -67,9 +82,12 @@ def calculate_coverage_score(cell: str, radius: int, df: pd.DataFrame) -> float:
     return total_risk * (1 + 0.1 * demand_weight)
 
 def build_prediction_grid(week: int, dow: int, hour: int, 
-                         demand_model, nulo_model) -> pd.DataFrame:
+                         demand_model, nulo_model, h3_resolution: int = 8) -> pd.DataFrame:
     """
-    Construye una cuadrícula de predicciones para todas las celdas H3.
+    Construye una cuadricula de predicciones para todas las celdas H3.
+    
+    Args:
+        h3_resolution: Resolución H3 (default=8 para áreas urbanas)
     """
     # Obtener todas las celdas H3 únicas de la base de datos
     from db import get_all_h3_cells
@@ -80,7 +98,13 @@ def build_prediction_grid(week: int, dow: int, hour: int,
     for h3_cell in h3_cells:
         try:
             # Convertir H3 a entero para el modelo
-            h3_int = int(h3_cell, 16)
+            # Asegurar que sea un string hexadecimal válido
+            if isinstance(h3_cell, str):
+                h3_int = int(h3_cell, 16)
+            else:
+                # Si ya es numérico o otro tipo
+                h3_str = str(h3_cell)
+                h3_int = int(h3_str, 16) if h3_str.startswith('8') else int(h3_str)
             
             # Predecir demanda
             X_demand = [[h3_int, week, dow, hour]]
@@ -98,7 +122,7 @@ def build_prediction_grid(week: int, dow: int, hour: int,
             coverage_weight = get_coverage_weight(expected_demand)
             
             predictions.append({
-                'h3': h3_cell,
+                'h3': str(h3_cell),  # Asegurar que es string
                 'h3_int': h3_int,
                 'week': week,
                 'dow': dow,
@@ -112,12 +136,20 @@ def build_prediction_grid(week: int, dow: int, hour: int,
             
         except Exception as e:
             print(f"Error procesando celda {h3_cell}: {e}")
+            if DEBUG_MODE:
+                import traceback
+                traceback.print_exc()
             continue
     
-    return pd.DataFrame(predictions)
+    df_result = pd.DataFrame(predictions)
+    
+    if df_result.empty:
+        print(f"Advertencia: No se generaron predicciones para week={week}, dow={dow}, hour={hour}")
+    
+    return df_result
 
 def place_gruas(df: pd.DataFrame, k: int = 2, 
-                max_iterations: int = 100) -> Dict:
+                max_iterations: int = 100, debug: bool = False) -> Dict:
     """
     Algoritmo de colocación de grúas greedy.
     
@@ -125,6 +157,7 @@ def place_gruas(df: pd.DataFrame, k: int = 2,
         df: DataFrame con predicciones por celda H3
         k: número de grúas a colocar
         max_iterations: máximo de iteraciones para evitar loops infinitos
+        debug: modo depuración para imprimir información
     
     Returns:
         Diccionario con las grúas seleccionadas y estadísticas
@@ -138,60 +171,87 @@ def place_gruas(df: pd.DataFrame, k: int = 2,
         if col not in df.columns:
             raise ValueError(f"DataFrame missing required column: {col}")
     
-    uncovered = df.copy()
+    # Hacer copia para trabajar
+    working_df = df.copy()
     selected_cells = []
     coverage_stats = []
     total_initial_risk = df['risk'].sum()
     
+    if debug:
+        print(f"Iniciando colocación de {k} grúas")
+        print(f"Total de celdas: {len(working_df)}")
+        print(f"Riesgo total inicial: {total_initial_risk:.4f}")
+    
     iteration = 0
-    while len(selected_cells) < k and iteration < max_iterations and not uncovered.empty:
+    while len(selected_cells) < k and iteration < max_iterations and not working_df.empty:
         iteration += 1
         best_cell = None
         best_score = -1
         best_coverage = set()
+        best_radius = 0
+        
+        if debug:
+            print(f"\nIteración {iteration}: Evaluando {len(working_df)} celdas candidatas")
         
         # Evaluar cada celda no cubierta como candidata
-        for _, row in uncovered.iterrows():
+        for idx, row in working_df.iterrows():
             current_cell = row['h3']
-            radius = row['coverage_radius']
+            radius = int(row['coverage_radius'])
             
             # Calcular celdas cubiertas
             covered = covered_cells(current_cell, radius)
             
             # Calcular puntaje de cobertura
-            score = calculate_coverage_score(current_cell, radius, uncovered)
+            score = calculate_coverage_score(current_cell, radius, working_df)
             
             if score > best_score:
                 best_score = score
                 best_cell = current_cell
                 best_coverage = covered
+                best_radius = radius
         
         if best_cell is None:
+            if debug:
+                print("No se encontró una mejor celda, terminando...")
             break
         
         # Añadir a seleccionadas
         selected_cells.append(best_cell)
         
+        if debug:
+            print(f"Seleccionada celda: {best_cell} con radio {best_radius}")
+            print(f"Puntaje: {best_score:.4f}, Celdas cubiertas: {len(best_coverage)}")
+        
+        # Obtener datos de la celda seleccionada
+        cell_data = df[df['h3'] == best_cell].iloc[0]
+        
         # Calcular estadísticas para esta grúa
         grua_stats = {
             'h3': best_cell,
-            'expected_demand': float(uncovered.loc[uncovered['h3'] == best_cell, 'expected_demand'].iloc[0]),
-            'coverage_radius': int(uncovered.loc[uncovered['h3'] == best_cell, 'coverage_radius'].iloc[0]),
-            'nulo_probability': float(uncovered.loc[uncovered['h3'] == best_cell, 'nulo_probability'].iloc[0]),
+            'expected_demand': float(cell_data['expected_demand']),
+            'coverage_radius': int(cell_data['coverage_radius']),
+            'nulo_probability': float(cell_data['nulo_probability']),
+            'risk': float(cell_data['risk']),
             'cells_covered': len(best_coverage),
-            'risk_covered': float(uncovered[uncovered['h3'].isin(best_coverage)]['risk'].sum())
+            'risk_covered': float(working_df[working_df['h3'].isin(best_coverage)]['risk'].sum())
         }
         coverage_stats.append(grua_stats)
         
         # Eliminar celdas cubiertas
-        uncovered = uncovered[~uncovered['h3'].isin(best_coverage)]
+        rows_before = len(working_df)
+        working_df = working_df[~working_df['h3'].isin(best_coverage)]
+        rows_after = len(working_df)
+        
+        if debug:
+            print(f"Celdas eliminadas: {rows_before - rows_after}, Quedan: {rows_after}")
     
-    # Calcular cobertura total
+    # Calcular cobertura total usando el DataFrame original
     covered_by_all = set()
     for cell in selected_cells:
-        cell_data = df[df['h3'] == cell].iloc[0]
-        covered = covered_cells(cell, cell_data['coverage_radius'])
-        covered_by_all.update(covered)
+        if cell in df['h3'].values:
+            cell_data = df[df['h3'] == cell].iloc[0]
+            covered = covered_cells(cell, int(cell_data['coverage_radius']))
+            covered_by_all.update(covered)
     
     total_cells_covered = len(covered_by_all)
     total_cells = len(df)
@@ -200,7 +260,7 @@ def place_gruas(df: pd.DataFrame, k: int = 2,
     total_risk_covered = df[df['h3'].isin(covered_by_all)]['risk'].sum()
     risk_coverage_percentage = (total_risk_covered / total_initial_risk * 100) if total_initial_risk > 0 else 0
     
-    return {
+    result = {
         'selected': selected_cells,
         'statistics': coverage_stats,
         'coverage': {
@@ -213,13 +273,24 @@ def place_gruas(df: pd.DataFrame, k: int = 2,
         },
         'parameters': {
             'k': k,
-            'iterations': iteration
+            'iterations': iteration,
+            'requested_k': k,
+            'actual_k': len(selected_cells)
         }
     }
+    
+    if debug:
+        print(f"\nResultado final:")
+        print(f"Grúas colocadas: {len(selected_cells)} de {k} solicitadas")
+        print(f"Cobertura: {coverage_percentage:.2f}% de celdas")
+        print(f"Cobertura de riesgo: {risk_coverage_percentage:.2f}%")
+    
+    return result
 
 def optimize_gruas_placement(df: pd.DataFrame, min_gruas: int = 1, 
                            max_gruas: int = 10, 
-                           target_coverage: float = 80.0) -> Dict:
+                           target_coverage: float = 80.0,
+                           debug: bool = False) -> Dict:
     """
     Encuentra el número óptimo de grúas para alcanzar una cobertura objetivo.
     
@@ -228,25 +299,45 @@ def optimize_gruas_placement(df: pd.DataFrame, min_gruas: int = 1,
         min_gruas: mínimo número de grúas
         max_gruas: máximo número de grúas
         target_coverage: porcentaje de cobertura objetivo
+        debug: modo depuración
     
     Returns:
         Resultado con número óptimo de grúas
     """
+    if df.empty:
+        return {"error": "No data available for optimization"}
+    
     results = []
     
+    if debug:
+        print(f"\nIniciando optimización:")
+        print(f"Rango de grúas: {min_gruas} a {max_gruas}")
+        print(f"Cobertura objetivo: {target_coverage}%")
+    
     for k in range(min_gruas, max_gruas + 1):
-        placement = place_gruas(df, k=k)
+        if debug:
+            print(f"\nProbando k = {k}...")
         
-        results.append({
+        placement = place_gruas(df, k=k, debug=debug)
+        
+        result_entry = {
             'k': k,
             'coverage_percentage': placement['coverage']['coverage_percentage'],
             'risk_coverage_percentage': placement['coverage']['risk_coverage_percentage'],
             'selected_cells': placement['selected'],
             'placement': placement
-        })
+        }
+        
+        results.append(result_entry)
+        
+        if debug:
+            print(f"  Cobertura: {result_entry['coverage_percentage']:.2f}%")
+            print(f"  Cobertura riesgo: {result_entry['risk_coverage_percentage']:.2f}%")
         
         # Si alcanzamos la cobertura objetivo, podemos detenernos
         if placement['coverage']['coverage_percentage'] >= target_coverage:
+            if debug:
+                print(f"¡Objetivo alcanzado! Cobertura: {placement['coverage']['coverage_percentage']:.2f}%")
             break
     
     # Encontrar el mejor balance
@@ -258,10 +349,58 @@ def optimize_gruas_placement(df: pd.DataFrame, min_gruas: int = 1,
     
     best = results[0]
     
+    # Verificar si el mejor resultado tiene suficiente cobertura
+    if best['coverage_percentage'] < 50 and debug:
+        print(f"Advertencia: La mejor solución solo cubre {best['coverage_percentage']:.2f}%")
+    
     return {
         'optimal_k': best['k'],
         'coverage_percentage': best['coverage_percentage'],
         'risk_coverage_percentage': best['risk_coverage_percentage'],
         'selected_cells': best['selected_cells'],
-        'all_results': results
-    }
\ No newline at end of file
+        'all_results': results[:10]  # Limitar a 10 resultados para no hacer la respuesta muy grande
+    }
+
+# Función auxiliar para detectar versión de H3
+def get_h3_version_info() -> Dict:
+    """Obtener información sobre la versión de H3 instalada."""
+    try:
+        import h3
+        version_info = {
+            'version': h3.__version__ if hasattr(h3, '__version__') else 'unknown',
+            'available_functions': [f for f in dir(h3) if not f.startswith('_')],
+            'api_type': 'standard'
+        }
+        
+        # Verificar funciones específicas
+        test_cell = "88390cb29dfffff"
+        test_radius = 1
+        
+        # Probar diferentes APIs
+        try:
+            # Probar API nueva
+            import h3.api.numpy_int as h3_new
+            cells = h3_new.grid_disk(test_cell, test_radius)
+            version_info['api_detected'] = 'numpy_int'
+            version_info['working_function'] = 'grid_disk'
+        except:
+            try:
+                # Probar k_ring directo
+                cells = h3.k_ring(test_cell, test_radius)
+                version_info['api_detected'] = 'standard_kring'
+                version_info['working_function'] = 'k_ring'
+            except AttributeError:
+                try:
+                    # Probar API antigua
+                    from h3 import h3 as h3_old
+                    cells = h3_old.k_ring(test_cell, test_radius)
+                    version_info['api_detected'] = 'old_api'
+                    version_info['working_function'] = 'h3.k_ring'
+                except:
+                    version_info['api_detected'] = 'unknown'
+                    version_info['working_function'] = 'none'
+        
+        return version_info
+        
+    except Exception as e:
+        return {'error': str(e), 'version': 'unknown'}
\ No newline at end of file