Chapter 10: Fleet Management and Routing

Learning Objectives

By the end of this chapter, you will be able to:

• Describe the delivery challenges that existed before wireless tank monitoring
• Explain how real-time tank data transforms fleet operations and routing decisions
• Apply route optimization algorithms to delivery scheduling problems
• Calculate the efficiency gains from data-driven delivery optimization
• Design dispatch workflows that integrate tank monitoring with fleet management
• Evaluate the TankScan claim of 30% delivery efficiency improvement

---

10.1 Introduction: The Last Mile of Liquid Logistics

Delivering liquid products to tanks is a "last mile" logistics problem. Regardless of how efficiently product is manufactured, refined, or transported in bulk, the final delivery to individual customer tanks determines the overall cost and customer experience. A fuel distributor with 2,000 customer tanks faces a daily puzzle: which tanks need product, how much should be delivered, and in what sequence should the driver visit them to minimize cost and maximize efficiency?

Before wireless monitoring, this puzzle was largely unsolvable because the distributor lacked the most critical piece of information: the current level in each customer's tank. Dispatchers worked from estimates, schedules, and customer phone calls. The result was inefficiency -- trucks running half-full, unnecessary trips to tanks that did not need product, and emergency deliveries to tanks that ran dry unexpectedly.

Wireless tank monitoring provides the missing data. When every tank reports its level multiple times per day, the dispatch puzzle transforms from guesswork into a solvable optimization problem.

The 30% Efficiency Claim

TankScan and other wireless monitoring providers frequently cite a 25-35% improvement in delivery fleet efficiency. This chapter examines the mathematical and operational basis for this claim, breaking it down into its component improvements and showing how each contributes to the total.

---

10.2 The Delivery Challenge (Before Wireless Monitoring)

10.2.1 Traditional Delivery Methods

Without real-time tank data, fuel and liquid product distributors relied on a combination of imperfect methods:

graph TD
    subgraph "Traditional Delivery Methods"
        A[Will-Call<br>Customer calls when<br>they need product] --> E[Dispatcher]
        B[Fixed Schedule<br>Deliver every N days<br>regardless of level] --> E
        C[Degree-Day Forecast<br>Estimate consumption<br>from weather data] --> E
        D[Driver Observation<br>Driver notes low<br>tanks during route] --> E

        E --> F[Build Route<br>Based on incomplete<br>information]
        F --> G[Execute Deliveries<br>Hope for the best]
    end

Each method had significant drawbacks:

Method	How It Works	Key Weakness	Impact
Will-call	Customer phones when tank is low	Late notice; often at crisis point	40-60% of deliveries are emergencies
Fixed schedule	Deliver every 14, 21, or 28 days	Ignores actual consumption	Over-delivery (tank still 60% full) or under-delivery (tank empty before schedule)
Degree-day	Estimate usage from temperature history	Assumes uniform housing/insulation	15-25% forecast error for individual tanks
Driver observation	Driver checks nearby tanks visually	Depends on route proximity; inconsistent	Misses tanks not on current route

10.2.2 The Cost of Uncertainty

Without knowing actual tank levels, distributors suffered predictable inefficiencies:

1. Partial loads (under-utilization)

When a dispatcher cannot predict exactly how much product each customer needs, they cannot fill the truck to capacity. Arriving at a tank that is 70% full means delivering only 30% of its capacity -- far less than the truck can carry. The truck returns to the terminal with unused capacity.

\[\text{Truck Utilization} = \frac{\text{Gallons Actually Delivered}}{\text{Truck Capacity}} \times 100\%\]

Typical pre-monitoring truck utilization: 55-65%

2. Unnecessary stops (wasted trips)

A fixed-schedule approach sends the truck to every customer on the schedule, regardless of whether they need product. If a customer's tank is still 75% full, the stop delivers minimal gallons but consumes the same travel time as a full delivery.

3. Emergency deliveries (crisis response)

When a tank runs out unexpectedly, the distributor must dispatch a truck immediately, regardless of the current route plan. Emergency deliveries are the most expensive type:

Delivery Type	Cost Multiple	Reason
Planned (on-route)	1.0x (baseline)	Included in optimized route
Unplanned (added to route)	1.5-2.0x	Route detour, partial load
Emergency (dedicated trip)	3.0-5.0x	Single destination, possibly overtime
After-hours emergency	5.0-8.0x	Overtime labor, on-call driver

4. Customer dissatisfaction

Run-outs cause immediate customer dissatisfaction. A home without heating fuel in winter, a gas station with empty pumps, or a construction site with idle equipment all result in lost revenue and damaged relationships.

10.2.3 Quantifying Pre-Monitoring Inefficiency

A representative mid-size fuel distributor before wireless monitoring:

Metric	Typical Value
Customer tanks served	1,500
Delivery trucks	8
Deliveries per day (fleet)	35-45
Average drop size	600-900 gallons
Truck utilization	55-65%
Gallons per mile	30-40
Emergency deliveries per month	25-40
Run-outs per month	5-15
Annual delivery cost per tank	$180-280

---

10.3 How Tank Data Transforms Fleet Operations

10.3.1 The Information Advantage

With wireless monitoring, the dispatcher has real-time access to the level, consumption rate, and days-to-empty for every tank in the fleet. This fundamentally changes the optimization problem:

graph LR
    subgraph "Before: Limited Information"
        A1[Schedule says<br>deliver to Customer A] --> B1[Drive to Customer A]
        B1 --> C1[Tank is 70% full<br>Deliver only 300 gal]
        C1 --> D1[Wasted trip capacity]
    end

    subgraph "After: Full Information"
        A2[AIP shows Customer A<br>at 70% - skip today] --> B2[AIP shows Customer B<br>at 15% - needs 2,800 gal]
        B2 --> C2[Deliver 2,800 gal<br>to Customer B]
        C2 --> D2[Efficient, full delivery]
    end

Key data points that enable optimization:

Data Point	How It Enables Optimization
Current level (%)	Know exactly which tanks need product
Days-to-empty (DTE)	Prioritize by urgency
Available capacity (ullage)	Know exactly how much to deliver
Consumption rate (gal/day)	Predict when non-urgent tanks will become urgent
Product type	Group same-product deliveries on the same truck
GPS location	Calculate optimal drive routes
Historical pattern	Predict future needs beyond current DTE

10.3.2 The Decision Framework

With complete tank data, the dispatcher's daily decision becomes a structured optimization:

graph TD
    A[Morning: AIP generates<br>delivery recommendations] --> B[Filter: Tanks below<br>reorder threshold]
    B --> C[Rank by urgency<br>DTE ascending]
    C --> D[Group by geography<br>Cluster nearby tanks]
    D --> E[Group by product<br>Same-product runs]
    E --> F[Assign to trucks<br>Match capacity to demand]
    F --> G[Optimize route sequence<br>Minimize total distance]
    G --> H[Generate driver manifests<br>with turn-by-turn directions]
    H --> I[Execute deliveries]
    I --> J[AIP confirms deliveries<br>via level increases]
    J --> K[Update forecasts<br>for tomorrow]

---

10.4 Route Optimization Algorithms

10.4.1 The Vehicle Routing Problem (VRP)

Delivery routing is a variant of the Vehicle Routing Problem (VRP), one of the most studied problems in operations research. The VRP asks: given a fleet of vehicles at a depot, a set of customers to serve, and constraints on vehicle capacity and driver hours, what is the optimal set of routes?

NP-Hard Complexity

The VRP is NP-hard, meaning there is no known algorithm that can find the absolute optimal solution in polynomial time for large problem instances. For a distributor with 200 tanks needing delivery, the number of possible route permutations exceeds the number of atoms in the universe. Practical solutions use heuristic and metaheuristic algorithms that find near-optimal solutions in reasonable computing time.

VRP formulation:

Minimize total route cost:

\[\text{Minimize } \sum_{k=1}^{K} \sum_{i=0}^{N} \sum_{j=0}^{N} c_{ij} \cdot x_{ijk}\]

Subject to constraints:

• Each customer is visited exactly once: \(\sum_{k=1}^{K} \sum_{i=0}^{N} x_{ijk} = 1 \quad \forall j\)
• Vehicle capacity: \(\sum_{j=1}^{N} d_j \cdot \sum_{i=0}^{N} x_{ijk} \leq Q_k \quad \forall k\)
• Route continuity: Each vehicle leaves every visited customer
• Driver hours: Total route time does not exceed shift length

Where:

• \(K\) = number of vehicles
• \(N\) = number of customers
• \(c_{ij}\) = cost of traveling from customer \(i\) to customer \(j\)
• \(x_{ijk}\) = 1 if vehicle \(k\) travels from \(i\) to \(j\), 0 otherwise
• \(d_j\) = demand at customer \(j\) (gallons needed)
• \(Q_k\) = capacity of vehicle \(k\)

10.4.2 Practical Optimization Approaches

Since exact VRP solutions are computationally infeasible for real-world problem sizes, practical systems use heuristic methods:

Algorithm	Approach	Quality	Speed	Best For
Nearest neighbor	Visit the closest unvisited customer next	Low-moderate	Very fast	Quick estimates, small fleets
Clarke-Wright savings	Merge routes that share nearby customers	Moderate-good	Fast	Medium-size problems
Sweep algorithm	Sweep a line from depot, group by angle	Moderate	Fast	Geographically distributed customers
Genetic algorithm	Evolve a population of candidate routes	Good-excellent	Moderate	Large, complex problems
Simulated annealing	Accept worse solutions to escape local optima	Good-excellent	Moderate	Complex constraints
Ant colony optimization	Simulate ant pheromone trails to find routes	Good	Moderate	Dynamic environments
Google OR-Tools	Open-source constraint solver	Excellent	Fast-moderate	Industry standard

10.4.3 Tank Monitoring-Specific VRP Extensions

Standard VRP assumes fixed customer demand. Tank monitoring adds several dimensions:

1. Variable demand (known through monitoring)

Unlike a package delivery where every customer gets one package, tank deliveries vary in size. Tank monitoring tells the dispatcher exactly how many gallons each tank needs:

\[d_j = V_{\text{capacity},j} \times \text{Fill Target \%} - V_{\text{current},j}\]

2. Time flexibility (DTE-based scheduling)

Not all deliveries are equally urgent. A tank with 2 days-to-empty must be served today; a tank with 10 days-to-empty can be deferred to a better route day.

\[\text{Urgency Score}_j = \frac{1}{\text{DTE}_j}\]

3. Product constraints (same-product routing)

A truck carrying diesel cannot simultaneously carry gasoline (unless it has compartments). Deliveries must be grouped by product type.

4. Delivery window constraints

Some customers have restricted delivery windows (e.g., gas stations may only accept fuel deliveries at night; residential customers may restrict delivery to business hours).

10.4.4 Optimization Example

graph TD
    subgraph "Before Optimization"
        A1[Terminal] --> B1[Customer A<br>3 miles, 200 gal]
        B1 --> C1[Customer B<br>8 miles, 400 gal]
        C1 --> D1[Customer C<br>12 miles, 300 gal]
        D1 --> E1[Customer D<br>6 miles, 500 gal]
        E1 --> F1[Terminal<br>Return]

        G1["Total: 29 miles<br>1,400 gal delivered<br>Truck: 3,000 gal capacity<br>Utilization: 47%"]
    end

    subgraph "After Optimization"
        A2[Terminal] --> B2[Customer D<br>4 miles, 500 gal]
        B2 --> C2[Customer E<br>2 miles, 800 gal]
        C2 --> D2[Customer A<br>5 miles, 600 gal]
        D2 --> E2[Customer F<br>3 miles, 700 gal]
        E2 --> F2[Customer B<br>4 miles, 400 gal]
        F2 --> G2[Terminal<br>5 miles]

        H2["Total: 23 miles<br>3,000 gal delivered<br>Truck: 3,000 gal capacity<br>Utilization: 100%"]
    end

The optimized route includes different customers (D, E, F instead of C, D) because the optimizer selected tanks that are emptier (needing more gallons) and closer together, resulting in more gallons delivered per mile.

---

10.5 Geographic Clustering

10.5.1 Cluster-Based Routing

Geographic clustering groups nearby customers into delivery zones, reducing the total distance traveled by keeping trucks within concentrated areas.

graph TD
    subgraph "Geographic Clustering"
        subgraph "Cluster 1: North Zone"
            A1[Tank A: 15%]
            A2[Tank B: 22%]
            A3[Tank C: 18%]
        end

        subgraph "Cluster 2: East Zone"
            B1[Tank D: 12%]
            B2[Tank E: 8%]
            B3[Tank F: 25%]
        end

        subgraph "Cluster 3: South Zone"
            C1[Tank G: 20%]
            C2[Tank H: 30%]
            C3[Tank I: 14%]
        end
    end

    D[Truck 1] --> A1 & A2 & A3
    E[Truck 2] --> B1 & B2 & B3
    F[Truck 3] --> C1 & C2 & C3

10.5.2 Clustering Algorithms

Algorithm	How It Works	Strengths	Weaknesses
K-means	Partition customers into K groups by proximity	Simple, fast	Assumes equal-size clusters
DBSCAN	Group by density, identify outliers	Handles irregular shapes	Sensitive to parameters
Hierarchical	Build tree of nested clusters	Flexible, visual	Computationally expensive
Grid-based	Divide service area into grid cells	Simple, predictable	Ignores road network
Road-network-based	Cluster by actual driving distance	Most realistic	Requires road network data

10.5.3 Dynamic Clustering with Tank Data

Static geographic clusters work well for fixed-schedule operations, but tank monitoring enables dynamic clustering -- regrouping based on which tanks actually need service each day.

Daily dynamic clustering workflow:

1. AIP identifies all tanks below reorder threshold (need delivery within N days)
2. Tanks are plotted on a map by GPS coordinates
3. Clustering algorithm groups nearby "needs-delivery" tanks
4. Each cluster is assigned to a truck
5. Within each cluster, route optimization determines the visit sequence

This dynamic approach means that clusters change daily based on actual tank levels. A customer who consumed more than expected joins today's cluster; one who consumed less drops out.

---

10.6 Delivery Scheduling Based on Tank Levels

10.6.1 The Scheduling Decision Matrix

Not every tank below the reorder point needs delivery today. The scheduling decision considers urgency, efficiency, and constraints:

Factor	Question	Impact on Scheduling
Days-to-empty	How urgent is this delivery?	Lower DTE = higher priority
Geographic proximity	Are there other nearby deliveries?	Cluster with others for efficiency
Drop size	How many gallons will be delivered?	Larger drops = more efficient
Product type	Can this tank share a truck load?	Same-product tanks grouped together
Delivery window	When can this site accept delivery?	Must fit within allowed times
Truck availability	Is a truck with the right product available?	Constrained by fleet composition
Weather	Will road/weather conditions allow access?	May delay rural/remote deliveries

10.6.2 Priority Scoring Algorithm

AIP uses a priority score to rank delivery candidates:

\[\text{Priority Score} = w_1 \cdot \frac{1}{\text{DTE}} + w_2 \cdot \frac{d_j}{Q_{\max}} + w_3 \cdot C_j + w_4 \cdot H_j\]

Where:

• \(\frac{1}{\text{DTE}}\) = urgency factor (higher when tank is closer to empty)
• \(\frac{d_j}{Q_{\max}}\) = efficiency factor (larger drops score higher)
• \(C_j\) = cluster bonus (tanks near other pending deliveries score higher)
• \(H_j\) = historical importance (tanks with history of run-outs score higher)
• \(w_1, w_2, w_3, w_4\) = configurable weights

Tuning Priority Weights

The default priority weights emphasize urgency (preventing run-outs). Distributors focused on efficiency can increase the efficiency weight (\(w_2\)) to favor larger drops. Those managing customer relationships can increase the cluster bonus (\(w_3\)) to favor routes with more stops in the same area.

10.6.3 Look-Ahead Scheduling

Instead of scheduling only for today, advanced systems use look-ahead scheduling to optimize over a multi-day horizon:

graph LR
    subgraph "3-Day Look-Ahead"
        A[Today<br>15 tanks need delivery] --> B[Tomorrow<br>8 more tanks will<br>cross threshold]
        B --> C[Day After<br>12 more tanks will<br>cross threshold]
    end

    A --> D{Can some of today's<br>deliveries wait 1-2 days<br>for a better route?}
    D -->|Yes| E[Defer to build<br>more efficient<br>future route]
    D -->|No| F[Deliver today<br>urgency too high]

Look-ahead scheduling can improve efficiency by 5-10% beyond single-day optimization. For example, if Customer A needs delivery today (DTE = 3) but Customer A's neighbor, Customer B, will need delivery tomorrow (DTE = 4), it may be more efficient to deliver both tomorrow rather than making two separate trips.

---

10.7 The 30% Efficiency Improvement: How It Works

10.7.1 Decomposing the Efficiency Gain

The commonly cited 25-35% improvement in delivery efficiency comes from multiple compounding factors:

Efficiency Lever	Individual Improvement	How It Works
Eliminate unnecessary stops	10-15%	Only visit tanks that actually need product
Increase average drop size	8-12%	Deliver when tanks are emptier (higher ullage)
Optimize route sequence	5-8%	Shorter total route distance
Reduce emergency deliveries	3-5%	Planned deliveries replace expensive emergencies
Better truck loading	3-5%	Know exact gallons needed; load truck optimally

Compounding effect:

These improvements compound. If you eliminate 12% of unnecessary stops AND increase drop size by 10% AND optimize routes by 6%, the combined effect is:

\[\text{Combined Improvement} = 1 - (1 - 0.12)(1 - 0.10)(1 - 0.06) = 1 - 0.744 = 25.6\%\]

10.7.2 Detailed Efficiency Calculations

Before monitoring:

Metric	Value
Total tanks served	1,500
Deliveries per year	12,000
Average drop size	750 gallons
Total gallons delivered/year	9,000,000
Total route miles/year	225,000
Gallons per mile	40
Delivery trucks	8
Cost per delivery	$85
Total delivery cost/year	$1,020,000

After monitoring (with optimization):

Metric	Value	Change
Total tanks served	1,500	Same
Deliveries per year	8,500	-29% (fewer, more productive stops)
Average drop size	1,060 gallons	+41% (tank is emptier when served)
Total gallons delivered/year	9,010,000	Same (demand unchanged)
Total route miles/year	157,000	-30%
Gallons per mile	57	+43%
Delivery trucks	6	-2 trucks (-25%)
Cost per delivery	$82	-4% (less time per stop)
Total delivery cost/year	$697,000	-32%

Annual savings: $323,000 (32% reduction)

Breaking Down the Savings

Savings Category	Annual Amount	Explanation
Fuel cost reduction	$68,000	68,000 fewer miles at $1.00/mile fuel cost
Vehicle maintenance	$34,000	Fewer miles = less wear, tires, oil changes
Driver labor	$120,000	Eliminated need for 2 drivers (including benefits)
Emergency delivery elimination	$48,000	30 fewer emergencies/month at $133 premium each
Vehicle lease/depreciation	$53,000	2 fewer trucks in fleet
Total	$323,000

10.7.3 Efficiency Metrics Comparison

graph TD
    subgraph "Key Efficiency Metrics"
        A[Gallons per Mile]
        B[Stops per Route]
        C[Gallons per Stop]
        D[Truck Utilization %]
        E[Emergency Rate]
    end

    A --> A1["Before: 40 gal/mi"]
    A --> A2["After: 57 gal/mi (+43%)"]

    B --> B1["Before: 5.2 stops"]
    B --> B2["After: 7.8 stops (+50%)"]

    C --> C1["Before: 750 gal"]
    C --> C2["After: 1,060 gal (+41%)"]

    D --> D1["Before: 58%"]
    D --> D2["After: 89% (+53%)"]

    E --> E1["Before: 8% of deliveries"]
    E --> E2["After: 1% of deliveries (-88%)"]

---

10.8 Dispatch Optimization

10.8.1 The Dispatcher's New Role

With wireless monitoring, the dispatcher's role shifts from firefighting to optimization:

Before Monitoring	After Monitoring
Answer customer calls requesting delivery	Review AIP's automated delivery recommendations
Guess which tanks need product	See exact tank levels and days-to-empty
Build routes from memory and experience	Optimize routes using algorithms and data
React to emergencies throughout the day	Plan proactively with minimal emergencies
Manually track deliveries via phone/radio	Track deliveries in real-time via GPS and auto-confirmation
Reconcile delivery tickets manually	Automated reconciliation via AIP delivery detection

10.8.2 Dispatch Workflow

sequenceDiagram
    participant AIP as AIP Platform
    participant Disp as Dispatcher
    participant Driver as Driver (Mobile App)
    participant Tank as Customer Tank

    Note over AIP: 5:00 AM - Daily planning run
    AIP->>AIP: Analyze all tank levels<br>Calculate DTE, priority scores
    AIP->>Disp: Delivery recommendations<br>35 tanks across 5 routes

    Note over Disp: 6:00 AM - Review and adjust
    Disp->>AIP: Accept Route 1-3, modify Route 4,<br>defer Route 5 to tomorrow
    AIP->>AIP: Re-optimize modified routes

    Note over Driver: 7:00 AM - Shift start
    AIP->>Driver: Route manifest:<br>8 stops, 6,800 gal, est. 7.5 hrs
    Driver->>Driver: Load truck at terminal

    Note over Driver: 7:30 AM - Route execution
    Driver->>Tank: Deliver to Stop 1 (1,200 gal)
    Tank->>AIP: Level increase detected
    AIP->>Disp: Delivery confirmed: Stop 1

    Driver->>Tank: Deliver to Stop 2 (850 gal)
    Tank->>AIP: Level increase detected
    AIP->>Disp: Delivery confirmed: Stop 2

    Note over Driver: Continue through all stops...

    Note over Driver: 3:30 PM - Route complete
    Driver->>AIP: Route complete: 8 stops, 6,650 gal
    AIP->>Disp: Route summary and variance report

10.8.3 Real-Time Route Adjustment

During route execution, conditions may change. The dispatch system should support real-time adjustments:

Situation	Response
Road closure or traffic	Reroute to alternate path
Customer unavailable	Skip stop, reschedule
Tank level changed (e.g., customer drew down more)	Adjust delivery quantity
New urgent delivery identified	Insert stop into current route if feasible
Driver running ahead of schedule	Add an opportunistic stop
Driver running behind schedule	Drop lowest-priority stop, reschedule
Truck mechanical issue	Transfer remaining stops to another truck

---

10.9 Driver Assignment and Load Optimization

10.9.1 Driver Assignment Factors

Assigning the right driver to each route considers multiple factors:

Factor	Consideration
CDL class	Truck type requires appropriate license class
Hazmat endorsement	Required for certain products
Hours of service (HOS)	DOT regulations limit driving and on-duty hours
Route familiarity	Experienced drivers on complex routes
Customer relationship	Some customers prefer specific drivers
Shift schedule	Match route duration to driver's shift
Equipment qualification	Driver must be trained on specific truck type

10.9.2 Load Optimization

Full truck vs. partial load:

The goal is to load each truck as close to capacity as possible for each route:

\[\text{Load Factor} = \frac{\sum_{j \in \text{Route}} d_j}{Q_k} \times 100\%\]

Where \(d_j\) is the demand at each stop on the route and \(Q_k\) is the truck capacity.

Load Factor	Assessment	Action
95-100%	Optimal	Execute route as planned
85-95%	Good	Acceptable; check for nearby add-on stops
70-85%	Moderate	Look for additional stops to improve utilization
< 70%	Poor	Combine with another partial route or defer stops

Multi-compartment trucks:

Many delivery trucks have multiple compartments for different products (e.g., a fuel truck with compartments for diesel and gasoline). Load optimization must consider:

\[\sum_{j \in \text{Route, Product P}} d_j \leq Q_{k,P}\]

Each product's total demand on the route must fit within the corresponding compartment.

Multi-Compartment Loading

A delivery truck has two compartments:

• Compartment 1: 3,500 gallons (diesel)
• Compartment 2: 2,500 gallons (gasoline)
• Total capacity: 6,000 gallons

Route has 10 stops:

Stop	Product	Gallons Needed
1	Diesel	800
2	Diesel	600
3	Gasoline	500
4	Diesel	450
5	Gasoline	700
6	Diesel	900
7	Gasoline	400
8	Diesel	500
9	Gasoline	600
10	Diesel	350

Diesel total: 3,600 gallons (exceeds 3,500 compartment capacity) Gasoline total: 2,200 gallons (under 2,500 compartment capacity)

Resolution: Defer Stop 10 (350 gal diesel, lowest priority) to tomorrow's route. Final load: Diesel: 3,250 gal (93%), Gasoline: 2,200 gal (88%), Total: 5,450 gal (91%)

---

10.10 Integration with Fleet Management Systems

10.10.1 System Integration Architecture

TankScan's AIP platform integrates with fleet management and dispatch systems to create a unified operational workflow:

graph TD
    subgraph "Tank Data Layer"
        A[TankScan AIP<br>Tank levels, DTE,<br>delivery recommendations]
    end

    subgraph "Fleet Management Layer"
        B[Fleet Management System<br>Omnitracs, Samsara, Geotab]
        C[GPS Tracking<br>Real-time vehicle location]
        D[ELD/HOS<br>Driver hours compliance]
        E[Vehicle Diagnostics<br>Maintenance alerts]
    end

    subgraph "Dispatch Layer"
        F[Dispatch Software<br>Route optimization,<br>driver assignment]
    end

    subgraph "Execution Layer"
        G[Driver Mobile App<br>Turn-by-turn, delivery confirmation]
    end

    subgraph "Business Layer"
        H[ERP / Accounting<br>Invoicing, inventory reconciliation]
    end

    A --> |API: Tank data| F
    B --> |API: Vehicle status| F
    C --> |API: Location data| F
    D --> |API: Available hours| F
    F --> |API: Route plan| G
    G --> |API: Delivery data| H
    A --> |API: Delivery confirmation| H

10.10.2 Integration Data Flows

Data Flow	Source	Destination	Frequency	Purpose
Tank levels and DTE	AIP	Dispatch software	Real-time	Delivery scheduling
Delivery recommendations	AIP	Dispatch software	Daily batch	Route planning
Vehicle location	GPS system	Dispatch software	Real-time	Route tracking
Driver hours available	ELD system	Dispatch software	Daily	Compliance
Route plan	Dispatch software	Driver mobile app	Per route	Execution
Delivery confirmation	AIP (level change detection)	ERP/accounting	Per delivery	Invoicing
Fuel consumption	Vehicle diagnostics	Fleet management	Continuous	Cost tracking

10.10.3 Common Integration Partners

System	Category	Integration Method	Key Data Exchanged
ADD Systems	Fuel distribution ERP	API, flat file	Orders, deliveries, invoices
Cargas Energy	Fuel distribution ERP	API	Delivery scheduling, invoicing
Omnitracs	Fleet management	API	Vehicle location, HOS
Samsara	Fleet management / telematics	API	GPS, vehicle health, HOS
Geotab	Fleet management / telematics	API	GPS, driver behavior
FuelCloud	Fuel management	API	Fuel dispensing data
P3 Software	Propane distribution	API, flat file	Delivery scheduling
Google Maps Platform	Mapping / routing	API	Geocoding, directions, traffic

---

10.11 GPS and Mapping Integration in AIP

10.11.1 Mapping Features

The AIP platform includes integrated mapping capabilities:

Feature	Description	Value
Tank location map	All monitored tanks plotted on a map	Visual fleet overview
Color-coded markers	Fill level indicated by marker color	Instantly identify low tanks
Cluster view	Zoom out to see geographic clusters	Identify delivery zones
Route visualization	Planned route drawn on map	Driver orientation
Geocoding	Automatic lat/lon from address	Accurate location data
Geofencing	Define geographic zones for alerts	Zone-based management
Traffic overlay	Real-time traffic data on routes	Realistic time estimates

10.11.2 Geographic Analysis for Fleet Planning

Tank location data enables strategic fleet planning:

Service area analysis:

Analysis	Question Answered	Planning Impact
Customer density map	Where are customers concentrated?	Terminal and depot placement
Drive time isochrones	How far can a truck travel in 4 hours?	Service area boundaries
Delivery frequency heat map	Where do trucks go most often?	Pre-positioning of product
Distance distribution	How far are customers from the terminal?	Fleet sizing and type selection
Growth analysis	Where are new customers being added?	Expansion planning

graph TD
    subgraph "Service Area Analysis"
        A[Terminal Location] --> B[1-hour drive radius<br>50-70 mile radius]
        B --> C[2-hour drive radius<br>100-140 mile radius]
        C --> D[3-hour drive radius<br>150-200 mile radius]

        B --> E["Core service area<br>80% of customers<br>Highest efficiency"]
        C --> F["Extended service area<br>15% of customers<br>Moderate efficiency"]
        D --> G["Remote service area<br>5% of customers<br>Consider satellite depot"]
    end

---

10.12 Real-World Efficiency Metrics

10.12.1 Industry Benchmarks

Based on published case studies and industry data, the following benchmarks represent typical performance with and without wireless monitoring:

Metric	Without Monitoring	With Monitoring	Industry Best
Gallons per mile	30-45	50-70	80+
Stops per route	4-7	7-12	14+
Average drop size (gal)	500-900	900-1,500	1,800+
Truck utilization	50-65%	80-95%	95%+
Emergency delivery rate	5-15%	0.5-2%	< 0.5%
Run-out rate (annual)	2-8%	0.1-0.5%	< 0.1%
Gallons per driver-hour	200-400	400-700	800+
Cost per gallon delivered	$0.06-0.12	$0.03-0.06	< $0.03
Annual deliveries per tank	8-15	5-8	4-6

10.12.2 Continuous Improvement Tracking

AIP provides dashboards for tracking efficiency metrics over time:

Efficiency Trend: Gallons per Mile

80 |
   |                                    *  *
70 |                              *  *
   |                        *  *
60 |                  *  *
   |            *  *
50 |      *  *
   |  *
40 |*
   |___________________________________
   Q1    Q2    Q3    Q4    Q1    Q2
   |<-- Year 1 -->|  |<-- Year 2 -->|

Note: Efficiency improves rapidly in first 6 months
as optimization algorithms learn consumption patterns,
then continues to improve gradually.

10.12.3 Factors That Limit Efficiency Gains

Not every deployment achieves the full 30% efficiency improvement. Factors that limit gains:

Factor	Impact	Mitigation
Customer density	Sparse, rural customers limit route optimization	Focus on cluster efficiency; consider depot placement
Product diversity	Many different products limit truck sharing	Multi-compartment trucks; product-specific routes
Delivery windows	Restricted delivery times reduce scheduling flexibility	Negotiate wider windows; prioritize restricted sites
Road network	Poor roads, seasonal access limit routing	Seasonal route planning; weather-based scheduling
Driver resistance	Drivers prefer familiar routes	Training; incentive programs tied to efficiency KPIs
Small fleet size	Fewer vehicles = less optimization flexibility	Focus on route sequencing and load optimization
Legacy systems	Inability to integrate dispatch and tank data	API integration; middleware solutions

---

10.13 Chapter Summary

Wireless tank monitoring transforms fleet management and delivery routing from a guessing game into a data-driven optimization problem. The key concepts covered in this chapter:

1. Traditional delivery methods (will-call, fixed schedule, degree-day) are fundamentally limited by the absence of real-time tank data
2. Tank monitoring provides the missing data -- current levels, consumption rates, and days-to-empty for every tank
3. Route optimization algorithms (VRP variants, geographic clustering) use this data to minimize delivery cost
4. The 30% efficiency improvement comes from compounding multiple factors: eliminating unnecessary stops, increasing drop sizes, optimizing route sequences, and reducing emergencies
5. Dispatch workflows shift from reactive firefighting to proactive optimization
6. Integration with fleet management systems creates an end-to-end operational platform
7. Real-world metrics consistently show significant improvements in gallons per mile, truck utilization, and delivery cost

The transformation is not merely incremental. It is a fundamental change in how liquid product distribution operates -- from a schedule-based, estimate-driven model to a demand-driven, data-optimized model. Organizations that fully embrace this transformation achieve sustainable competitive advantages in cost, service quality, and customer retention.

---

Review Questions

Question 1 -- Knowledge (Remember)

List the five main efficiency levers that contribute to the 30% delivery efficiency improvement claim, and state the approximate individual contribution of each.

Question 2 -- Comprehension (Understand)

Explain the difference between static geographic clustering and dynamic clustering based on tank monitoring data. Describe a scenario where dynamic clustering would produce significantly better results than static clustering.

Question 3 -- Application (Apply)

A fuel distributor has a truck with 5,000-gallon capacity and the following tanks needing delivery today:

Tank	Location	Product	Gallons Needed	DTE
A	North zone	Diesel	1,200	2
B	North zone	Diesel	800	5
C	East zone	Diesel	1,500	3
D	East zone	Gasoline	900	1
E	South zone	Diesel	600	7
F	North zone	Diesel	1,100	4

The truck can only carry one product type per trip. Build the optimal diesel delivery route, specifying which tanks to include, the load quantity, the estimated truck utilization, and which tank(s) to defer. Justify your choices.

Question 4 -- Analysis (Analyze)

A distributor deployed wireless monitoring on 1,000 tanks six months ago. They expected a 30% efficiency improvement but have achieved only 12%. Analyze the possible reasons for this gap. Consider factors related to: (a) data quality and system configuration, (b) dispatch process adoption, (c) driver behavior, (d) customer and geographic characteristics, and (e) integration with existing systems. For each factor, suggest a diagnostic test and a corrective action.

Question 5 -- Evaluation (Evaluate)

Compare the following two delivery strategies for a propane distributor serving 3,000 residential customers:

Strategy A: Monitor all 3,000 tanks with TSD gauges, use AI-driven dispatch optimization, deliver only when tanks reach 20%.

Strategy B: Monitor only the top 500 highest-volume customers with TSD gauges, use degree-day forecasting for the remaining 2,500, deliver when estimated at 30%.

Evaluate both strategies on: (a) investment cost, (b) expected efficiency improvement, (c) run-out risk, (d) customer equity concerns, and (e) long-term competitiveness. Which would you recommend and why?