Modelling the Logistics of Mantzikert

Cover
Title Page
Copyright page
Contents
List of Figures
	Figure 1: Byzantine Anatolia at the time of the Turkish raids.
	Figure 2: Estimates of the size of the Byzantine army at Mantzikert from Muslim sources.
	Figure 3: Locations mentioned in the Byzantine historical accounts of the campaign.
	Figure 4: Resource availability in Byzantine Anatolia (after Hendy 2008, 70).
	Figure 5: Different types of pack saddle (Furse 1882, 280).
	Figure 6: Number of military logistics publications in the timeline, divided by date range.
	Figure 7: The software used in the MWGrid system.
	Figure 8: The same point of a simulation rendered with both lower polygon (left) and higher polygon (right) models.
	Figure 9: A 2D representation of MWGrid agents in a typical camp layout.
	Figure 10: A 3D representation of MWGrid agents moving across the landscape. The two lines of yellow represent columns of infantry agents making their way to the night’s camp, at which a blob of blue and purple cavalry have already arrived.
	Figure 11: An example of A* in action.
	Figure 12: Grid Movement.
	Figure 13: PRM movement.
	Figure 14: Anatolia with the ABM terrain extent marked in red.
	Figure 15: Layout of Byzantine camp from the Treatise on Campaign Organisation and Tactics (after Dennis, 1985) (image by Nigel Dodds).
	Figure 16: Example of camp layout in the DM101 scenarios, with 101 agents in red, 10,001 agents in blue and 40,001 agents in grey.
	Figure 17: Arrival tick of the Column Leader in marches of differing distance over flat terrain.
	Figure 18: Arrival time of the last agent in marches of differing distance over flat terrain.
	Figure 19: DM102 arrival times for 100% infantry (blue) and 75% infantry (green).
	Figure 20: DM102 arrival times for 90% infantry (yellow) and 50% infantry (black).
	Figure 21: DM102 arrival times for all scenarios.
	Figure 22: DM103 10km arrival times.
	Figure 23: DM103 20km arrival times.
	Figure 24: DM103 30km arrival times.
	Figure 25: Order of march for two columns, marching southeast. Red sections travel via waypoint 1 in numerical order and yellow sections via waypoint 2.
	Figure 26: Order of march for three columns travelling southeast. Red section travels via waypoint one, yellow sections via waypoint two and grey via waypoint three.
	Figure 27: The three parallel columns with the corresponding waypoints outside the starting and destination camps. 
	Figure 28: DM106 arrival time for 10,001 agents over 10km.
	Figure 29: DM107 arrival times for armies with 508 mule squads over 10km.
	Figure 30: DM107 arrival times for armies with 1428 mule squads over 10km.
	Figure 31: DM107 arrival times for only agents who completed the day’s march, with 4118 mule squads over 10km.
	Figure 32: The hypothetical route of the Byzantine army to Mantzikert.
	Figure 33: Start and end point of the two DM108 marches based around Leukai and Dorylaion.
	Figure 34: DM108 height profiles of marches from Leukai and Dorylaion using 1, 2 or 3 columns.
	Figure 35: DM108 relationship between distance covered and kilocalories expended.
	Figure 36: The size of the hypothetical army as it progresses towards Mantzikert.
	Figure 37: Route map from Helenopolis to Nikaia.
	Figure 38: Arrival tick of units on the march from Helenopolis to Nikaia - Day 1.
	Figure 39: Route map from Nikaia to Dorylaion.
	Figure 40: Arrival time of units on the march between Nikaia and Leukai - Day 1.
	Figure 41: Cross section of the first day’s march between Nikaia and Leukai.
	Figure 42: Unit arrival time on march between Leukai and Bilecik.
	Figure 43: Route map between Dorylaion and Charsianon.
	Figure 44: Unit arrival time for march between Dorylaion and Charsianon.
	Figure 45: Route map between Charsianon and Sebastea.
	Figure 46: Route map between Sebastea and Theodosiopolis.
	Figure 47: Route map between Theodosiopolis and Mantzikert.
	Figure 48: Comparison of arrival times between DM107 & DM112 baggage setup with food for 1 day.
	Figure 49: Comparison of arrival times between DM107 & DM112 baggage setup with food for 10 days.
	Figure 50: Comparison of arrival times between DM107 & DM112 baggage setup with food for 20 days.
	Figure 51: Arrival times of units with (blue) and without (orange) the emperor’s baggage train.
	Figure 52: Pie chart showing army composition in individuals, with matching colours to Figure 53
	Figure 53: Visualisation of camp layout, with agent types coloured as in Figure 52
List of Tables
	Table 1: Size of the Byzantine army from Arabic sources
	Table 2: A summary of the maximum loads of various pack animals
	Table 3: A summary of water requirements for pack animals
	Table 4: A summary of food requirements for pack animals
	Table 5: The implications of Nikephoros Ouranos’ recommended total of arrows on cavalry size
	Table 6: A timeline of selected military logistics writing
	Table 7: Aggregated statistical output of a day’s march
	Table 8: Simulation characteristics of pack animals
	Table 9: A* first planning move
	Table 10: A* second planning move
	Table 11: Effects of heuristic values on the running of the A* algorithm
	Table 12: Arrival tick and travel time of the Column Leader in marches of differing distance over flat terrain
	Table 13: Number of agents still on the march at the end of a 12 hour simulation over flat terrain
	Table 14: Travel speed of the different agent types
	Table 15: DM102 aggregate data for different army compositions
	Table 16: DM103 scenario labelling
	Table 17: DM103 10km arrival and travel times
	Table 18: DM103 20km arrival and travel times
	Table 19: DM103 30km arrival and travel times
	Table 20: DM106 agents not arrived by the end of the day’s march
	Table 21: DM106 aggregate stats for 10,001 agents over 10km
	Table 22: DM107 aggregate stats
	Table 23: The location of settlements along the route to Mantzikert within the ABMs co-ordinate system
	Table 24: Kilocalories consumed while horse riding (after Devienne and Guezennec, 2000)
	Table 25: DM108 aggregate stats
	Table 26: DM111 distances and army composition along the hypothetical route
	Table 27: Sunrise and sunset data used in DM111
	Table 28: Simulation data from a day’s march between Dorylaion and Charsianon
	Table 29: Arrival data for half the army - Sebastea to Theodosiopolis
	Table 30: Arrival data for the full army - Sebastea to Theodosiopolis
	Table 31: Quantities and organisation of mules in the DM112 system
	Table 32: DM112 and DM107 aggregate stats with enough food for 1 day
	Table 33: DM112 and DM107 aggregate stats with enough food for 10 days
	Table 34: DM112 and DM107 aggregate stats with enough food for 20 days
	Table 35: Aggregate data from two scenarios: ‘NoEmp’ without the emperor’s baggage train and ‘Emp’ with.
	Table 36: Average weights of animal products from cattle and sheep
	Table 37: Meat on the hoof requirements for our simulated army.
	Table 38: Number of individuals in DM112 hypothetical army
	Table 39: DM112 squads marching from Theodosiopolis with siege machinery and cattle
Acknowledgements
Chapter 1 – Introduction
	What is military logistics?
	The importance of logistics
	Mantzikert - our case study
	The historical context
	The importance of Mantzikert
	The campaign
	What is missing?
	The need to model
	Why can this not be filled with conventional research?
	The model-based approach
	The problem of complexity
	A brief history of ABM
	ABM in archaeology
	Military historical simulation
	Is the Byzantine army a complex system?
	The limitations of ABM
	The potential of ABM
	Summary
	A note on names and dates
Chapter 2 – The Historical Context
	Introduction
	Byzantine and contemporary sources for the Mantzikert campaign
	Relevant sources for the Byzantine army in other situations
	Military treatises
	Conclusion
Chapter 3 - The Byzantine World
	Byzantine taxation
	Byzantine agriculture
	Byzantine demography
	The Byzantine road network
	Summary
Chapter 4 - Logistical Considerations
	Military rations
	Water
	Climate
	Animals and loads
	Equipment
	Camps
	Health and hygiene
	Routine on the march
	Non-combatants
Chapter 5 – 19th-Century Military Writing
	Introduction
	The problem with prior work
	19th century military writing
	Comparing Byzantine and modern military sources
	Examples of marches from other works
	Logistics failures
	Modern developments
	Conclusion – where does this leave us?
Chapter 6 – The Models
	Introduction
	Is modelling valid?
	The complete MWGrid software suite
	The ABM
	Data processing with OpenOffice
	Data processing with Blender
	Formats of visualisation
	Agents
	The environment
	Route planning
	A* route planning
	What is the solution?
	Verification and validation
	Exploring the parameter space
Chapter 7 – Results
	Introduction
	Units of measurement
	Initial setup
	The Day’s March scenarios
	DM101 – Size and distance
	DM102 – Composition
	DM103 – Resting and squad spacing
	DM106 – Splitting the army into separate columns
	DM107 – Baggage
	DM108 – Terrain and calories consumed
	DM111 – The Mantzikert campaign
	DM112 – Supply systems
Chapter 8 – Conclusions
	Synthesis of results
	Advantages over previous work
	Simulating Furse
	Reassessing Engels
	Individual variation
	Biographies of simulated people
	Technical achievements
	Advanced technological considerations
	Future work
	Concluding remarks
Bibliography