AVEVA PI Web API Integration

Overview

Enable Claude to retrieve historian data from the Boeing AVEVA PI System through the PI Web API, supporting current value queries, historical time-series retrieval, and Asset Framework navigation to discover PI Points.

Quick Start

Get Current Value by Tag Name

from scripts.pi_client import PIWebAPIClient
import os

client = PIWebAPIClient(
    username=os.environ["PI_USERNAME"],
    password=os.environ["PI_PASSWORD"]
)

# Get current value
value = client.get_current_value_by_tag("TAG001")
print(f"{value['Value']} {value['UnitsAbbreviation']}")

Get Historical Data

# Last 24 hours of recorded data
data = client.get_recorded_values_by_tag(
    tag_name="TAG001",
    start_time="*-24h",
    end_time="*",
    max_count=1000
)

for item in data['Items']:
    print(f"{item['Timestamp']}: {item['Value']}")

Navigate AF to Find PI Points

# Get element
element = client.get_element_by_path(
    "\\\\AF_SERVER\\ProductionData\\Site1\\Machine123"
)

# Get attributes (PI Points)
attributes = client.get_element_attributes(element['WebId'])

# Find PI Point attributes
for attr in attributes['Items']:
    if attr.get('DataReferencePlugIn') == 'PI Point':
        value = client.get_value_by_webid(attr['WebId'])
        print(f"{attr['Name']}: {value['Value']}")

Core Capabilities

1. Current Value Retrieval

Get the most recent value for a PI tag.

Methods:

• get_current_value_by_tag(tag_name) - By tag name
• get_value_by_webid(webid) - By WebId (faster if cached)

Use when:

• Need latest sensor reading
• Checking current equipment status
• Real-time monitoring

Example:

value = client.get_current_value_by_tag("MACHINE123.TEMP")
if value['Good']:
    print(f"Temperature: {value['Value']}°F")
else:
    print("WARNING: Bad quality data")

2. Historical Data Retrieval

Query time-series data over specified ranges.

Data Types:

• Recorded - Archive values as stored in PI
• Interpolated - Regular interval samples
• Plot - Optimized for visualization
• Summary - Aggregated statistics (min, max, avg)

Time expressions:

• Relative: *-1d (1 day ago), *-8h (8 hours ago)
• Absolute: 2024-01-15T00:00:00Z
• Special: * (now), t (today), y (yesterday)

Methods:

# Recorded values (as archived)
data = client.get_recorded_values_by_tag(
    "TAG001",
    start_time="*-7d",
    end_time="*",
    max_count=10000
)

# Interpolated (hourly samples)
data = client.get_interpolated_values_by_tag(
    "TAG001",
    start_time="*-24h",
    end_time="*",
    interval="1h"
)

# Summary statistics (daily averages)
data = client.get_summary_values_by_tag(
    "TAG001",
    start_time="*-30d",
    end_time="*",
    summary_type="Average",
    summary_duration="1d"
)

See references/examples.md for more time range patterns.

3. Asset Framework Navigation

Drill down through AF hierarchy to find PI Points.

Workflow:

1. Get database by path
2. Navigate elements (Site → Area → Equipment → Machine)
3. Get element attributes
4. Filter for PI Point data references
5. Retrieve stream data

Example:

# Get database
db = client.get_asset_database_by_path("\\\\AF_SERVER\\ProductionData")

# Get root elements
roots = client.get_elements(db['WebId'])

# Navigate to machine
machine = client.get_element_by_path(
    "\\\\AF_SERVER\\ProductionData\\Site1\\Area1\\Machine123"
)

# Get attributes with PI Point references
attributes = client.get_element_attributes(machine['WebId'])
pi_points = [
    attr for attr in attributes['Items']
    if attr.get('DataReferencePlugIn') == 'PI Point'
]

# Get current values for all PI Points
for attr in pi_points:
    value = client.get_value_by_webid(attr['WebId'])
    print(f"{attr['Name']}: {value['Value']} {value.get('UnitsAbbreviation', '')}")

4. Point Search

Find PI tags using wildcard patterns.

Use when:

• User doesn't know exact tag name
• Finding all tags for a machine
• Discovering available points

Example:

# Find all tags for MACHINE123
points = client.search_points(name_filter="MACHINE123.*")

for point in points['Items']:
    print(f"{point['Name']}: {point.get('Descriptor', '')}")

Configuration

Authentication

PI Web API uses HTTP Basic Authentication.

Setup:

# Environment variables
export PI_USERNAME="your_username"
export PI_PASSWORD="your_password"

Client initialization:

client = PIWebAPIClient(
    base_url="https://PI1AVDEVA.web.boeing.com/piwebapi",
    username=os.environ["PI_USERNAME"],
    password=os.environ["PI_PASSWORD"],
    default_data_server="PI1AVDEVA"
)

Config File Pattern

# config.json
{
  "pi_webapi": {
    "base_url": "https://PI1AVDEVA.web.boeing.com/piwebapi",
    "username": "username",
    "password": "password",
    "default_data_server": "PI1AVDEVA"
  }
}

# Load config
import json
with open('config.json') as f:
    config = json.load(f)

client = PIWebAPIClient(**config['pi_webapi'])

Common Patterns

Pattern 1: Get Machine Data

User asks: "What's the current temperature for Machine 123?"

# Option 1: Direct tag query (if you know the tag)
value = client.get_current_value_by_tag("MACHINE123.TEMP")

# Option 2: Navigate AF (if tag is unknown)
machine = client.get_element_by_path(
    "\\\\AF_SERVER\\Production\\Site1\\Machine123"
)
attributes = client.get_element_attributes(machine['WebId'])
temp_attr = next(
    attr for attr in attributes['Items']
    if 'temp' in attr['Name'].lower()
)
value = client.get_value_by_webid(temp_attr['WebId'])

print(f"Temperature: {value['Value']}°F")

Pattern 2: Time-Series Analysis

User asks: "Show me the pressure trend for the last 24 hours"

# Get hourly samples for smooth visualization
data = client.get_interpolated_values_by_tag(
    "MACHINE123.PRESSURE",
    start_time="*-24h",
    end_time="*",
    interval="1h"
)

# Extract values for analysis
timestamps = [item['Timestamp'] for item in data['Items']]
values = [item['Value'] for item in data['Items']]

# Could then plot or analyze trend

Pattern 3: Data Quality Checks

Always validate data quality before using values:

value = client.get_current_value_by_tag("TAG001")

if not value.get('Good', False):
    print(f"WARNING: Data quality issue")
    print(f"  Questionable: {value.get('Questionable', False)}")
    print(f"  Substituted: {value.get('Substituted', False)}")
else:
    # Use value
    print(f"Valid value: {value['Value']}")

Pattern 4: WebId Caching

WebIds are persistent - cache them for performance:

# Store WebIds to avoid repeated path lookups
webid_cache = {}

def get_value_cached(tag_name):
    if tag_name not in webid_cache:
        point = client.get_point_by_path(f"\\\\PI1AVDEVA\\{tag_name}")
        webid_cache[tag_name] = point['WebId']
    
    return client.get_value_by_webid(webid_cache[tag_name])

# First call: queries path and caches WebId
value1 = get_value_cached("TAG001")

# Second call: uses cached WebId (faster)
value2 = get_value_cached("TAG001")

Error Handling

Common errors:

• 401: Invalid credentials
• 404: Tag/path not found
• 400: Invalid time expression or parameters
• 403: Insufficient permissions

Defensive pattern:

try:
    value = client.get_current_value_by_tag("TAG001")
    print(f"Value: {value['Value']}")
except ValueError as e:
    if "404" in str(e):
        print("Tag not found")
    elif "401" in str(e):
        print("Authentication failed")
    else:
        print(f"Error: {e}")

Response Structure

Value Response

{
    "Timestamp": "2024-01-15T14:30:00Z",
    "Value": 75.3,
    "UnitsAbbreviation": "°F",
    "Good": True,
    "Questionable": False,
    "Substituted": False
}

Time-Series Response

{
    "Items": [
        {"Timestamp": "...", "Value": 72.5, "Good": True},
        {"Timestamp": "...", "Value": 73.8, "Good": True}
    ],
    "UnitsAbbreviation": "°F"
}

Element Response

{
    "WebId": "F1ABC...",
    "Name": "Machine123",
    "Path": "\\\\AF_SERVER\\DB\\Site\\Machine123",
    "HasChildren": True,
    "Links": {"Elements": "...", "Attributes": "..."}
}

Attribute Response

{
    "WebId": "F1DEF...",
    "Name": "Temperature",
    "Type": "Double",
    "DataReferencePlugIn": "PI Point",  # Indicates PI Point reference
    "ConfigString": "\\\\PI1AVDEVA\\TAG001",
    "DefaultUnitsName": "degree Fahrenheit"
}

Best Practices

1. Always specify time ranges - Avoid open-ended queries
2. Set maxCount limits - Default 1000, adjust as needed
3. Check data quality - Use Good, Questionable, Substituted flags
4. Cache WebIds - They're persistent and faster than path lookups
5. Use appropriate data type:
   • Recorded: Raw archive data
   • Interpolated: Regular intervals for analysis
   • Plot: Visualization (most efficient)
   • Summary: Statistics (min, max, avg)
6. Handle time zones - API returns UTC, convert as needed
7. Prefer WebId over path - Direct WebId lookups are faster

Resources

• references/api_reference.md - Complete endpoint documentation
• references/examples.md - Comprehensive usage examples
• scripts/pi_client.py - Production-ready client implementation

External Resources

• PI Web API Documentation: https://docs.aveva.com/bundle/pi-web-api-reference
• Boeing PI Server: https://PI1AVDEVA.web.boeing.com/piwebapi
• Default Data Server: PI1AVDEVA