app.py

import streamlit as st
import pandas as pd
import numpy as np
import joblib
import math

# ============================================================
# PAGE CONFIG
# ============================================================

st.set_page_config(
    page_title="HTRI Heat Exchanger Predictor",
    page_icon="🔥",
    layout="wide"
)

# ============================================================
# LOAD MODEL BUNDLE
# ============================================================

@st.cache_resource
def load_models():
    bundle = joblib.load("htri_models_v4.pkl")
    return bundle

try:
    bundle       = load_models()
    model_multi  = bundle["model_multi"]
    model_length = bundle["model_length"]
    model_qty    = bundle["model_qty"]
except Exception as e:
    st.error(f"Failed to load model: {e}")
    st.stop()

# ============================================================
# ROUNDING HELPERS
# ============================================================

def round_to_nearest(value, multiple):
    """Round value to nearest multiple."""
    return round(round(value / multiple) * multiple, 10)

def round_shell_od(value_in):
    """Round Shell OD to nearest 2 inches."""
    return int(round_to_nearest(value_in, 2))

def round_tube_od(value_in):
    """Round Tube OD to nearest 1/8 inch."""
    return round(round_to_nearest(value_in, 0.125), 4)

def round_tube_length(value_ft):
    """Round Tube Length to nearest whole foot."""
    return int(round(value_ft))

# ============================================================
# HEADER
# ============================================================

st.title("🔥 HTRI Heat Exchanger Geometry Predictor")

st.markdown(
    "Predict shell-and-tube heat exchanger geometry using a "
    "trained XGBoost ML model. Enter process conditions and "
    "get predicted sizing outputs instantly."
)

st.divider()

# ============================================================
# UNIT TOGGLE
# ============================================================

col_toggle, _ = st.columns([1, 3])

with col_toggle:
    unit_system = st.radio(
        "Input Unit System",
        options=["US Customary", "SI / Metric"],
        horizontal=True
    )

si_mode = unit_system == "SI / Metric"

st.divider()

# ============================================================
# INPUT FORM
# ============================================================

st.subheader("Process Inputs")

col1, col2, col3 = st.columns(3)

# ============================================================
# COLUMN 1 — FLOW RATES + HEAT DUTY
# ============================================================

with col1:

    st.markdown("**Flow Rates**")

    if si_mode:
        shell_flow = st.number_input(
            "Shell Side Flow Rate (kg/hr)",
            min_value=0.0, value=166000.0, step=100.0,
            format="%.1f"
        )
        tube_flow = st.number_input(
            "Tube Side Flow Rate (kg/hr)",
            min_value=0.0, value=112000.0, step=100.0,
            format="%.1f"
        )
    else:
        shell_flow = st.number_input(
            "Shell Side Flow Rate (lb/hr)",
            min_value=0.0, value=366000.0, step=100.0,
            format="%.1f"
        )
        tube_flow = st.number_input(
            "Tube Side Flow Rate (lb/hr)",
            min_value=0.0, value=247000.0, step=100.0,
            format="%.1f"
        )

    st.markdown("**Heat Duty**")

    if si_mode:
        heat_duty = st.number_input(
            "Heat Duty (kW)",
            min_value=0.0, value=5570.0, step=10.0,
            format="%.1f"
        )
    else:
        heat_duty = st.number_input(
            "Heat Duty (Btu/hr)",
            min_value=0.0, value=19000000.0, step=10000.0,
            format="%.1f"
        )

# ============================================================
# COLUMN 2 — THERMAL + PASSES + TUBE GEOMETRY
# ============================================================

with col2:

    st.markdown("**Thermal**")

    if si_mode:
        lmtd = st.number_input(
            "Corrected LMTD (°C)",
            min_value=0.0, value=109.4, step=0.1,
            format="%.2f"
        )
    else:
        lmtd = st.number_input(
            "Corrected LMTD (°F)",
            min_value=0.0, value=196.9, step=0.1,
            format="%.2f"
        )

    st.markdown("**Passes**")

    shell_passes = st.number_input(
        "Shell Passes",
        min_value=1, max_value=6, value=1, step=1
    )

    tube_passes = st.number_input(
        "Tube Passes",
        min_value=1, max_value=16, value=4, step=1
    )

    st.markdown("**Tube Geometry**")

    if si_mode:
        tube_pitch = st.number_input(
            "Tube Pitch (mm)",
            min_value=1.0, value=25.4000, step=0.0001,
            format="%.4f"
        )
    else:
        tube_pitch = st.number_input(
            "Tube Pitch (inch)",
            min_value=0.1, value=1.0000, step=0.0001,
            format="%.4f"
        )

    tube_layout = st.selectbox(
        "Tube Layout Angle (°)",
        options=[30, 45, 60, 90],
        index=3
    )

# ============================================================
# COLUMN 3 — PRESSURE DROPS + VELOCITIES
# ============================================================

with col3:

    st.markdown("**Pressure Drops**")

    if si_mode:
        shell_dp = st.number_input(
            "Shell Side ΔP (kPa)",
            min_value=0.0, value=3.63, step=0.01,
            format="%.3f"
        )
        tube_dp = st.number_input(
            "Tube Side ΔP (kPa)",
            min_value=0.0, value=6.87, step=0.01,
            format="%.3f"
        )
    else:
        shell_dp = st.number_input(
            "Shell Side ΔP (psi)",
            min_value=0.0, value=0.526, step=0.001,
            format="%.3f"
        )
        tube_dp = st.number_input(
            "Tube Side ΔP (psi)",
            min_value=0.0, value=0.996, step=0.001,
            format="%.3f"
        )

    st.markdown("**Velocities**")

    if si_mode:
        shell_vel = st.number_input(
            "Shell Side Velocity (m/s)",
            min_value=0.0, value=0.616, step=0.001,
            format="%.3f"
        )
        tube_vel = st.number_input(
            "Tube Side Velocity (m/s)",
            min_value=0.0, value=0.594, step=0.001,
            format="%.3f"
        )
    else:
        shell_vel = st.number_input(
            "Shell Side Velocity (ft/s)",
            min_value=0.0, value=2.020, step=0.001,
            format="%.3f"
        )
        tube_vel = st.number_input(
            "Tube Side Velocity (ft/s)",
            min_value=0.0, value=1.950, step=0.001,
            format="%.3f"
        )

st.divider()

# ============================================================
# CONVERT ALL INPUTS TO US CUSTOMARY FOR MODEL
# ============================================================

if si_mode:
    shell_flow_us = shell_flow * 2.20462
    tube_flow_us  = tube_flow  * 2.20462
    heat_duty_us  = heat_duty  * 3412.14
    lmtd_us       = (lmtd * 1.8) + 32
    tube_pitch_us = tube_pitch / 25.4
    shell_dp_us   = shell_dp   * 0.145038
    tube_dp_us    = tube_dp    * 0.145038
    shell_vel_us  = shell_vel  * 3.28084
    tube_vel_us   = tube_vel   * 3.28084
else:
    shell_flow_us = shell_flow
    tube_flow_us  = tube_flow
    heat_duty_us  = heat_duty
    lmtd_us       = lmtd
    tube_pitch_us = tube_pitch
    shell_dp_us   = shell_dp
    tube_dp_us    = tube_dp
    shell_vel_us  = shell_vel
    tube_vel_us   = tube_vel

# ============================================================
# PREDICT BUTTON
# ============================================================

predict = st.button("🚀 Predict Geometry", type="primary", use_container_width=True)

if predict:

    # --------------------------------------------------------
    # BUILD INPUT DATAFRAME
    # --------------------------------------------------------

    input_data = pd.DataFrame([{
        "Shell_Flow_lb_hr"   : shell_flow_us,
        "Tube_Flow_lb_hr"    : tube_flow_us,
        "Heat_Duty_Btu_hr"   : heat_duty_us,
        "LMTD_F"             : lmtd_us,
        "Shell_Passes"       : shell_passes,
        "Tube_Passes"        : tube_passes,
        "Tube_Pitch_in"      : tube_pitch_us,
        "Tube_Layout_Angle"  : tube_layout,
        "Shell_DP_psi"       : shell_dp_us,
        "Tube_DP_psi"        : tube_dp_us,
        "Shell_Velocity_ft_s": shell_vel_us,
        "Tube_Velocity_ft_s" : tube_vel_us,
    }])

    # --------------------------------------------------------
    # STEP 1 — Predict Shell_OD + Tube_OD (MultiOutput)
    # --------------------------------------------------------

    multi_pred  = model_multi.predict(input_data)
    shell_od_raw = float(multi_pred[0, 0])
    tube_od_raw  = float(multi_pred[0, 1])

    # --------------------------------------------------------
    # STEP 2 — Build chained input (raw values for model)
    # --------------------------------------------------------

    input_chained = input_data.copy()
    input_chained["Shell_OD_in"] = shell_od_raw
    input_chained["Tube_OD_in"]  = tube_od_raw

    # --------------------------------------------------------
    # STEP 3 — Predict Tube_Length + Tube_Quantity
    # --------------------------------------------------------

    tube_length_raw = float(model_length.predict(input_chained)[0])
    tube_qty_raw    = float(model_qty.predict(input_chained)[0])

    # --------------------------------------------------------
    # APPLY ROUNDING RULES
    # Shell OD  -> nearest 2 in
    # Tube OD   -> nearest 1/8 in
    # Tube Length -> nearest whole foot
    # Tube Qty  -> nearest whole number
    # --------------------------------------------------------

    shell_od_in  = round_shell_od(shell_od_raw)       # int, multiple of 2
    tube_od_in   = round_tube_od(tube_od_raw)          # float, multiple of 0.125
    tube_len_ft  = round_tube_length(tube_length_raw)  # int, whole feet
    tube_qty     = int(round(tube_qty_raw))            # int

    # SI conversions of rounded values
    shell_od_mm  = round(shell_od_in * 25.4, 1)
    tube_od_mm   = round(tube_od_in  * 25.4, 4)
    tube_len_m   = round(tube_len_ft * 0.3048, 3)

    # ±15% bounds
    shell_od_lo  = round_shell_od(shell_od_raw * 0.85)
    shell_od_hi  = round_shell_od(shell_od_raw * 1.15)
    tube_len_lo  = round_tube_length(tube_length_raw * 0.85)
    tube_len_hi  = round_tube_length(tube_length_raw * 1.15)
    tube_qty_lo  = int(round(tube_qty_raw * 0.85))
    tube_qty_hi  = int(round(tube_qty_raw * 1.15))

    # --------------------------------------------------------
    # RESULTS
    # --------------------------------------------------------

    st.subheader("📊 Predicted Geometry")

    # ---- High confidence ----
    st.markdown("#### ✅ High Confidence")
    st.caption("R² > 0.89 — reliable for preliminary sizing")

    r1, r2 = st.columns(2)

    with r1:
        if si_mode:
            st.metric(
                "Shell OD",
                f"{shell_od_mm} mm",
                delta=f"Raw: {round(shell_od_raw * 25.4, 1)} mm → rounded to nearest 50.8 mm (2 in)"
            )
        else:
            st.metric(
                "Shell OD",
                f"{shell_od_in} in",
                delta=f"Raw: {round(shell_od_raw, 3)} in → rounded to nearest 2 in"
            )

    with r2:
        if si_mode:
            st.metric(
                "Tube OD",
                f"{tube_od_mm} mm",
                delta=f"Raw: {round(tube_od_raw * 25.4, 4)} mm → rounded to nearest 3.175 mm (1/8 in)"
            )
        else:
            st.metric(
                "Tube OD",
                f"{tube_od_in} in",
                delta=f"Raw: {round(tube_od_raw, 4)} in → rounded to nearest 1/8 in"
            )

    st.divider()

    # ---- Moderate confidence ----
    st.markdown("#### ⚠️ Moderate Confidence")

    st.warning(
        "**Tube Length and Tube Quantity carry ±15% uncertainty** "
        "based on model validation (R² ≈ 0.79–0.82). "
        "Use these as a preliminary starting point and verify "
        "against your required heat transfer area."
    )

    r3, r4 = st.columns(2)

    with r3:
        if si_mode:
            st.metric(
                "Tube Length",
                f"{tube_len_m} m  ({tube_len_ft} ft)",
                delta=f"±15%  →  {round(tube_len_lo * 0.3048, 2)} – {round(tube_len_hi * 0.3048, 2)} m"
            )
        else:
            st.metric(
                "Tube Length",
                f"{tube_len_ft} ft",
                delta=f"±15%  →  {tube_len_lo} – {tube_len_hi} ft"
            )

    with r4:
        st.metric(
            "Tube Quantity",
            f"{tube_qty} tubes",
            delta=f"±15%  →  {tube_qty_lo} – {tube_qty_hi} tubes"
        )

    st.divider()

    # ---- Full summary table ----
    st.markdown("#### 📋 Full Summary")

    summary = pd.DataFrame({
        "Output": [
            "Shell OD",
            "Tube OD",
            "Tube Length",
            "Tube Quantity",
        ],
        "US Customary": [
            f"{shell_od_in} in",
            f"{tube_od_in} in",
            f"{tube_len_ft} ft",
            f"{tube_qty}",
        ],
        "SI / Metric": [
            f"{shell_od_mm} mm",
            f"{tube_od_mm} mm",
            f"{tube_len_m} m",
            f"{tube_qty}",
        ],
        "Rounding Rule": [
            "Nearest 2 in (50.8 mm)",
            "Nearest 1/8 in (3.175 mm)",
            "Nearest whole foot",
            "Nearest whole number",
        ],
        "Confidence": [
            "✅ High  (R² = 0.90)",
            "✅ High  (R² = 0.977)",
            "⚠️ Moderate  (R² = 0.79, ±15%)",
            "⚠️ Moderate  (R² = 0.82, ±15%)",
        ],
    })

    st.dataframe(summary, use_container_width=True, hide_index=True)

# ============================================================
# FOOTER
# ============================================================

st.divider()

st.caption(
    "Model: XGBoost Hybrid v4 | Trained on 1,507 HTRI datasheets | "
    "For preliminary sizing only — not a substitute for rigorous thermal design."
)