BIM in assessing the suitability of historic fortifications for military purposes, using the example of the gunpowder laboratory at Boyen Fortress in Giżycko

Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

doi:10.37190/arc260308

Abstract

Building Information Modeling (BIM) is increasingly used to document, analyse and protect historic buildings, enabling the creation of digital models that can serve as the basis for a variety of spatial analyses. For fortifications, the possibility of using such models to assess potential suitability in defence scenarios is particularly significant. Using the gunpowder laboratory at Boyen Fortress in Giżycko, Poland, as an example, the article presents a simplified BIM model as a tool for analysing the building’s relationship with its surrounding terrain and exploring conceptual alternatives for terrain interventions such as trenches.

The digital model of the building and its surroundings, supplemented by spatial analyses, enabled the assessment of spatial conditions, identification of potential threat directions and comparison of possible locations and appearances of earthworks. Because the BIM environment has limitations in modelling complex terrain geometry at the conceptual stage, generative artificial intelligence tools were used to visualise alternatives for potential earthworks.

The results indicate that even simplified BIM models can provide a useful basis for comparing design alternatives and documenting historic buildings. Artificial intelligence tools are primarily applicable in the early stages of conceptual work, supporting decision-making by quickly presenting possible solutions while retaining their illustrative character.

Download article

Full article view is only available on bigger screens.

Full article

2026

3(87)

Andrzej Szymon Borkowski*, Klaudia Zelkowska**,

Małgorzata Żach***

BIM in assessing the suitability of historic fortifications

for military purposes, using the example of the gunpowder laboratory

at Boyen Fortress in Giżycko

DOI: 10.37190/arc260308

Published in open access. CC BY-NC license.

Abstract

Building Information Modelling (BIM) is nding increasingly widespread application in the documentation, analysis and conservation of historic

buildings, enabling the creation of digital models which can then be used as a basis for a variety of spatial analyses. In the context of fortications, the

ability to use such models to assess their potential suitability in defence scenarios is of particular signicance. This article uses the example of the gun-

powder laboratory at Boyen Fortress in Giżycko (Poland) to illustrate the application of a simplied BIM model as a tool to support the analysis of the

structure’s relationship with the surrounding terrain and the conceptual evaluation of various terrain modications, such as trenches. A digital model of

the structure and its surroundings, supplemented by spatial analyses, enabled the assessment of spatial conditions, the identication of potential threat

directions, and the comparison of possible solutions for the location and appearance of earthworks. Due to the limitations of the BIM environment in

modelling complex terrain geometry at the conceptual stage, generative articial intelligence tools were used to visualise variants of potential earth-

works. The results obtained indicate that even simplied BIM models can provide a useful basis for variant analyses and the documentation of historic

structures. At the same time, articial intelligence-based tools are primarily used in the early stages of conceptual work, supporting the decision-making

process by rapidly presenting possible solutions whilst maintaining their illustrative nature.

Key words: BIM, Boyen Fortress, point cloud, trench, articial intelligence, gunpowder laboratory

Introduction

The authors’ focus is on a highly signicant example of

Prussian fortications – Boyen Fortress, situated in Gi życ-

ko, in the north-eastern part of Poland, in the Warmian-Ma-

surian Voivodeship (Fig. 1a) – and on methods of digital

surveying using Building Information Modelling (BIM)

technology. The fortress is situated between two large lakes

(Niegocin and Kisajno) and is one of the region’s most im-

portant historic sites (Krahel 2003). The gunpowder labo-

ratory, which is the subject of detailed research, is located

in the eastern part of the complex (Fig. 1b). In the 19

-cen-

tury functional layout of the fortress, this building served

as a site for the preparation of explosives and ammunition,

and its function diered from that of the military gunpow-

der store, which at Boyen Fortress was situated in the court-

yard of the Światło Bastion. The laboratory is characterised

by a distinctive design that can be described as being dug

into the ground; that is, from a bird’s-eye view, only the

corridor leading to the nine rooms is visible. This design

served, amongst other things, to protect the fortress’s other

structures in the event of an explosion (Twierdza Boyen

Gi życko n.d.).

Historic sites, such as Boyen Fortress, should be digi-

tised for their own preservation, given their cultural value.

BIM enables the creation of digital models of various his-

toric structures that reect their actual structure and charac-

ORCID:

0000-0002-7013-670X.

Faculty of Geodesy and Car -

tography, Warsaw University of Technology, Poland, e-mail: andrzej.

borkowski@pw.edu.pl

ORCID:

0009-0003-5226-0859.

Faculty of Geodesy and Car-

tography, Warsaw University of Technology, Poland.

***

ORCID:

0009-0000-2397-0366.

Faculty of Geodesy and Car-

tography, Warsaw University of Technology, Poland.

80 Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

teristics. Such models serve not only as documentation of

the current state of preservation, but also as a tool facilitat-

ing subsequent renovation, revitalisation or reconstruction

in the event of damage caused by human activity or natural

forces (Borkowski, Olszewska 2024). Creating a model us-

ing BIM preserves knowledge of the structure’s appearance

and condition, which, in the event of damage, would al-

low for the reconstruction of a historically valuable build-

ing, but also enable an assessment of the suitability of the

now-listed structure for contemporary military and defence

purposes. There are many dierent ways in which existing

historical infrastructure can be utilised for these purposes.

One of these is to design a trench as a simulation of poten-

tial defensive operations during wartime.

The aim of the authors is to present a methodology for

using a simplied BIM model, combining photogram-

metric data from unmanned aerial vehicles (UAVs) with

a variant analysis of site interventions supported by gen-

erative articial intelligence (AI) tools, in relation to his-

toric fortication structures. The case study focuses on the

gunpowder magazine at Boyen Fortress, for which digital

documentation of the external condition and an analysis

of illustrative conceptual intervention options have been

proposed, within the context of the debate on the state’s

defence readiness, whilst giving absolute priority to the

protection of the historic structure.

It is not the authors’ intention to advocate the active

military use of historic fortications. Nor does this paper

constitute an encouragement to undertake interventions that

would alter the substance of the historic monument. The

authors maintain that the protection of cultural heritage re-

mains the overriding priority. Any considerations regarding

the potential use of historic sites in crisis scenarios are pure-

ly of a hypothetical planning nature.

The BIM digital model fulls a dual role in this study.

Firstly, it serves as documentation of the site’s state of pres-

ervation. This function is purely conservation-related and

is independent of any further analyses. Secondly, the model

serves as a tool for a preliminary, conceptual spatial assess-

ment of possible on-site interventions. The second function

takes on signicance in the context of the ongoing debate in

Poland regarding the state’s defence readiness and in light

of the Regulation of the Minister of the Interior and Ad-

ministration of 2025 on protective and concealed structures

(Rozporządzenie… 2025).

It should be emphasised that variant analysis in a digital

environment is, by its very nature, intended to minimise the

risk of actual, irreversible interventions in a historic monu-

ment. It allows for the assessment of alternatives and their

consequences even before any on-site actions are under-

taken. In this sense, the use of BIM does not conict with

the mission of heritage conservation, but rather supports

it. It provides conservationists, defence services and deci-

sion-makers with a risk assessment tool.

The innovative contribution of this work does not lie

in the UAV data acquisition technology itself, nor in the

conversion of point clouds into a Revit model. These pro-

cedures are well documented in the relevant literature (Bor-

kowski, Kubrat 2024). The original contribution consists

of three elements. The rst of these is the integration of

variant analysis of conceptual site interventions with the

BIM model of the historic fortication. The second ele

ment

Fig. 1. Location of Boyen Fortress: a) its position in relation to the administrative boundaries of Poland and its provinces (elaborated by M. Żach,

K. Zelkowska based on the State Boundary Register of the General Directorate for Geodesy, Cartography and Cadastre [GUGiK];

https://www.geoportal.gov.pl/), b) location of the gunpowder laboratory in relation to the boundaries of Boyen Fortress (elaborated by M. Żach,

K. Zelkowska based on the GUGiK orthophoto map; https://www.geoportal.gov.pl/)

Il. 1. Lokalizacja Twierdzy Boyen: a) położenie na tle granic administracyjnych Polski i województw (oprac. M. Żach, K. Zelkowska

na podstawie Państwowego Rejestru Granic GUGiK; https://www.geoportal.gov.pl/), b) położenie laboratorium prochowego na tle granic

Twierdzy Boyen (oprac. M. Żach, K. Zelkowska na podstawie ortofotomapy GUGiK; https://www.geoportal.gov.pl/)

 BIMinassessingthesuitabilityofhistoricforticationsformilitarypurposes 81

is the use of generative articial intelligence tools, such as

Gemini 2.5 Flash Image (Nano Banana Pro) and GPT. These

tools were utilised to supplement the BIM environment in

areas where it exhibits signicant limitations. This primarily

concerns the modelling of complex terrain geometry in Re-

vit. The third element is a critical analysis of the limitations

of this hybrid methodology in conservation applications.

State of research

The eld of digital documentation of historic buildings

within the BIM environment has been developing for near-

ly two decades. Heritage Building Information Modelling

(HBIM) was introduced by Maurice Murphy and his co-au-

thors (Murphy, McGovern and Pavia 2009) as an extension

of the BIM methodology to include historic buildings, whose

geometry and construction do not conform to standard com-

ponent libraries. Subsequent work has focused on develop-

ing libraries of parametric elements for historic architecture

(Dore, Murphy 2017), the integration of data from terrestrial

laser scanning and photogrammetry (López et al. 2018), as

well as the challenges associated with the level of detail and

reliability of historical models (Penjor et al. 2024).

The number of studies on the application of BIM and

HBIM in the documentation of historic buildings is steadi-

ly increasing; however, the issue of using simplied BIM

models to assess the potential suitability of such buildings

in the context of modern defence remains a poorly explored

topic. In particular, issues combining spatial analysis based

on such a model with conceptual variations in the use of

existing terrain and structures are rarely addressed.

Methods

Unmanned aerial vehicles are now a widely used and

cost-eective tool for acquiring high-resolution data, as,

compared to methods based on, for example, satellite data,

they oer the possibility of acquiring data quickly and at

relatively low cost (Liu et al. 2022). This study utilised

a point cloud of the Boyen Fortress gunpowder laborato-

ry acquired during a ight with a DJI Mini 3 drone. Ap-

proximately 300 high-resolution images were captured and

subsequently processed photogrammetrically to produce

a three-dimensional point cloud depicting the gunpowder

laboratory and its immediate surroundings. The resulting

average resolution (GSD, Ground Sampling Distance) was

1.71 cm/pixel, which enabled a high-quality representation

of the object to be obtained. Following processing of the

point cloud in Agisoft software, a point cloud comprising

5,876,161 points was obtained. Due to the absence of an

RTK (Real-Time Kinematic) module in the drone used, the

point cloud was positioned in the WGS84 coordinate sys-

tem with a total error of 1.197 m.

The data processing began by importing the photo-

graphs into Agisoft and performing the image alignment

stage using the Align Photos tool. In this step, the software

searched for common points between the photographs,

identifying control points, and then calculated the orien-

tation and position of the camera for each shot. On this

basis, a preliminary, low-resolution point cloud was gen-

erated, which served as the basis for further processing. In

addition, during the scan of the terrain, photopoints were

utilised, which also served as an aid during the photo align-

ment process. In the next step, a dense point cloud was gen-

erated for further processing.

The resulting point cloud was imported into Autodesk

ReCap, where it underwent initial renement. At this stage,

a transformation to the local reference system (0,0,0) was

carried out, as working with geodetic coordinates can cause

reading and performance issues within the Autodesk Revit

environment. At the same time, the point cloud was man-

ually cleaned by removing noise irrelevant to the analysis,

in particular vegetation such as trees. After cleaning, the

point cloud was exported to .e57 format and imported into

CloudCompare.

In CloudCompare, the orientation of the point cloud was

corrected by correctly setting the reference plane. The

point

cloud was then exported in two formats: LAS and as an ASCII

text le. The LAS le was re-imported into Autodesk Re-

Cap and converted to the .rcp format, suitable for use in

Auto desk Revit. The point cloud prepared in this way was

imported into Revit using the “centre-to-centre” insertion

setting. An ASCII text le containing the point coordinates,

which were used to create the terrain model, was also im-

ported into Revit. It is important to maintain the correct le

structure and to set the separators to commas when import-

ing in CloudCompare, so that the point coordinates are cor-

rectly recognised during import.

Based on the imported point cloud, a simplied model

of the gunpowder laboratory was created, using the point

cloud as a measurement reference for determining the ele-

vations and geometry of the walls and ceiling. The model

was then rened to include basic elements such as doors

and windows, and materials were assigned to key compo-

nents. It should be emphasised, however, that the model is

for illustrative purposes only and does not fully capture the

details, as the photographic data was obtained from a bird’s-

eye view, which limits the accuracy of the representation of

vertical elements and façade details. The model was visu-

alised using Revit and Lumion. After integrating the mod-

elled building with the site, an integrated BIM model was

produced, serving as the basis for further analysis; the next

stage involves developing an excavation variant within the

modelled structure. During the development of the trench

simulation visualisations, two articial intelligence-based

tools were used to generate and manipulate images.

The trench visualisations were created in the ChatGPT

environment using the GPT model (OpenAI) and in Gemi-

ni 2.5 Flash Image, commonly known as Nano Banana Pro

(Google). The individual stages of the methodology em-

ployed, covering data acquisition, data processing, the de-

velopment of the BIM model and the generation of visualisa-

tions of the variants, are summarised in a owchart (Fig. 2).

Trenches as a defensive structure

Trenches are earthworks constituting an anthropogen-

ic feature of the landscape (Franczak, Jucha 2015). They

enable relatively safe movement during shelling, as well

as the conduct of combat operations. The method of con-

82 Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

In this paper, the trench is treated as an exemplary, hy-

pothetical site intervention, serving as a test case to assess

the suitability of a simplied BIM model for the conceptu-

al variation of defensive solutions and for evaluating their

spatial and conservation implications.

Results

The simplied model of the gunpowder laboratory

(Fig. 3a–d) reects the external geometry of the building,

the topography of the surrounding moat and the relation-

ship between the building’s massing and ground level.

However, it is important to clearly highlight the method-

ological limitations of the representation. The photogram-

metric data were acquired exclusively from a drone. By its

very nature, this precludes direct mapping of the interior of

the structure, in particular the corridor leading to the nine

laboratory rooms. The underground section of the building

remains outside the range of the point cloud obtained from

a UAV ight without LiDAR. This is a signicant limita-

tion, as it is precisely this part of the building that is key

to its historical function. However, this limitation is inher-

structing trenches depends on many dierent factors, such

as the type of subsoil, the topography, the anticipated con-

struction time and the available labour. Trenches are often

constructed in the form of a zigzagging trench, which pro-

vides protection in the event that the enemy penetrates the

interior of the trench (Twierdza Włocławek 2025). At the

same time, the method of constructing trenches is regulated

by applicable regulations, which specify their basic form,

structure and permissible dimensions. In accordance with

the Regulation of the Minister of the Interior and Admini-

stration, trenches are constructed as narrow excavations with

a broken plan, with or without a cover, with temporary shor-

ing made, for example, of wooden formwork or sandbags, or

without shoring, whilst maintaining the natural slope

angle.

The regulations also specify the permissible dimensions of

trenches, which, in the case of shored solutions, include

a height of 150–220 cm and a width ranging from approx-

imately 80–100 cm at the base to 100–120 cm at the top of

the covering, whilst for trenches without formwork or cov-

ering, the height is approximately 140 cm and the width is

greater at the top, resulting from the use of the natural slope

angle (Rozporządzenie… 2025).

Fig. 2. Diagram of creating a model from a point cloud using photographs obtained by flying a DJI Mini drone (elaborated by M. Żach, K. Zelkowska)

Il. 2. Schemat utworzenia modelu z chmury punktów ze zdjęć pozyskanych w wyniku nalotu dronem DJI Mini (oprac. M. Żach, K. Zelkowska)

 BIMinassessingthesuitabilityofhistoricforticationsformilitarypurposes 83

ent to the methodology adopted. A complete representation

of the interior would require supplementary ground-based

laser scanning (TLS) and interior scanning. The model

presented in this paper serves as a building-terrain model

rather than a full HBIM model. This is a deliberate choice.

The cross-sections shown in Figure 3c therefore illustrate

the relationship between the building’s structure and the

surrounding moat, rather than its internal functional lay-

out. A key outcome of the work carried out is the ability

to use the developed BIM model as digital documentation

of the building’s external condition. It records the geome-

try of the gunpowder laboratory and its relationship with

the surrounding terrain at the time the data was collected;

thus, in the event of partial or total destruction caused by

military action, natural disasters or deterioration over time,

the model can facilitate renovation or reconstruction work.

As part of the assessment of the military potential of the

Boyen Fortress gunpowder laboratory, simulations were

carried out to explore the possibility of transforming the

site by creating a visualisation of a trench. A variant anal-

ysis was conducted on the immediate surroundings of the

facility, taking into account the terrain, existing greenery,

spatial conditions and possible attack scenarios.

A threat scenario has been adopted in which the direction

of a potential attack is from the north-east (Fig. 4). From

the selected alternative location of the trench, the Boyen

Fortress area directly adjoins the built-up area of Giżycko,

Fig. 4. Visualisation of the attack scenario against the background

of an orthophotomap (elaborated by M. Żach, K. Zelkowska

based on an orthophotomap;

https://www.geoportal.gov.pl/)

Il. 4. Wizualizacja scenariusza ataku na tle ortofotomapy

(oprac. M. Żach, K. Zelkowska na podstawie ortofotomapy;

https://www.geoportal.gov.pl/)

Fig. 3. Model of the gunpowder laboratory at Boyen Fortress: a) topography, b) interior of the building, c) cross-section in Revit,

d) model of the Boyen Fortress gunpowder laboratory in Giżycko, including the topography (elaborated by M. Żach, K. Zelkowska)

Il. 3. Model laboratorium prochowego Twierdzy Boyen: a) ukształtowanie terenu, b) wewnętrzna część budynku, c) przekrój w programie Revit,

d) model laboratorium prochowego Twierdzy Boyen w Giżycku wraz z ukształtowaniem terenu (oprac. M. Żach, K. Zelkowska)

84 Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

making this direction one of the most accessible in terms of

transport links and potentially the most vulnerable to an ad-

vance by enemy forces. The scenario assumed that, follow-

ing the capture of the town, the Boyen Fortress area could

serve as one of the last points of resistance. Furthermore,

a key factor inuencing the choice of the trench’s location

is the natural terrain barriers surrounding the fortress. To

the east and north-west, the fortress area lies in close prox-

imity to large lakes (Niegocin and Kisajno). These signi-

cantly restrict direct access to the fort and hinder the con-

duct of land manoeuvres. From the south-west, a potential

advance towards the fortress is possible, though it is also

hampered by the presence of smaller lakes, such as Tajty,

Duże Wilkasy and Małe Wilkasy, which constitute addi-

tional natural obstacles. At the same time, from a broader

regional perspective, the northern and north-eastern direc-

tions are areas from which, geographically speaking, forc-

es posing a potential threat from beyond Poland’s borders

could move into Polish territory; in a hypothetical threat

scenario, this justies the potential location of the trench.

Furthermore, it should be noted that, from the southern side,

despite the presence of terrain obstacles, communication is

possible; and in the assumed siege scenario, the transport

routes located there could serve as a logistical base and fa-

cilitate supplies to the fortress area.

The appearance of trenches – that is, their form, shape

and length – depends on many factors, the key ones being

the terrain conditions, the time available for construction

and their intended function. Trenches can full several ba-

sic functions, including combat, defence and communica-

tion. However, two types of trenches can be distinguished:

ring trenches, which mark the front lines, and communi-

cation trenches, which connect them (Doyle 2019). As part

of the variant analysis carried out, two potential locations

for a trench were identied in the immediate vicinity of

the gunpowder laboratory building. Using articial intelli-

gence and based on the attached .jpg image of the gunpow-

der laboratory model (Fig. 3a), two variants of the trench’s

location and appearance were generated. The process of

generating the visualisations was based on descriptive text

prompts, which took into account the trench’s alignment,

its relationship with the terrain, the proximity to the gun-

powder laboratory building, and general characteristics

of earthworks, such as a zigzag shape or the shielding of

positions. The generated visualisations are for illustrative

purposes only. They serve to illustrate the spatial dierenc-

es between the variants. The geometric parameters of the

trenches adopted at the stage of generating prompts for the

AI tools were consistent with the dimensions specied in

the Regulation of the Minister of the Interior and Adminis-

tration of 2025. This applies to heights of 150–220 cm and

widths of 80–120 cm for the shored variants (Rozporządze-

nie… 2025). It should be noted, however, that the accu-

racy of dimension representation in images generated by

diusion models remains limited. These tools capture pro-

portions and spatial character, but should not be treated as

technical documentation meeting formal requirements. Full

detailed design drawings for the fortications would need

to be developed in a separate, dedicated design process, in

accordance with military regulations and guidelines.

The rst variant (Fig. 5a) involves the construction of

a long trench with a zigzagging and highly winding course,

situated in the upper part of the site, in the immediate vi-

cinity of the gunpowder laboratory, on the eastern side.

The trench’s course makes use of the natural topography.

Its length allows for the creation of a continuous line of

defence and the establishment of a front line, as well as the

deployment of several ring positions. The trench can also

serve a communication function, as it allows the garrison to

move relatively safely within the trench. It is also partially

extended closer to the gunpowder laboratory building, thus

providing better access to its rooms. The sharply zigzag-

ging shape of the trench limits the possibility of anking

re and reduces the eects of a shock wave (Doyle 2019).

This variant requires a greater amount of earthworks.

The second variant (Fig. 5b) involves constructing

a trench

that is also zigzag-shaped, but with a less winding

course. It is also situated in the upper part of the site, but

is shorter in length. The limited length of the fortication

means it serves a more combat-oriented function and can

act as an observation and cover position. The smaller num-

ber of bends and the shorter length facilitate quicker con-

struction and reduce the scope of earthworks. This makes

this variant a more viable option where construction time

is limited. At the same time, its defensive potential is more

limited compared with the rst variant.

A comparison of the options highlights, above all, the dif-

fering nature of the proposed solutions and the associated

design trade-os. The rst option oers greater potential in

terms of territorial defence and transport functionality, which

is its main advantage in the context of protracted hostilities.

At the same time, this solution involves a much greater need

for terrain modication and more demanding construction

requirements. Option two is characterised by a shorter con-

struction time, which may be its key advantage in situations

requiring the immediate securing of the area. However, the

reduced length and simpler layout of the fortication result

in a lower level of protection and reduced functionality. This

comparison shows that the proposed variants do not compete

directly with one another, but can be utilised in various situ-

ations where the priorities may be either to maximise defen-

sive potential or to minimise implementation time and inter-

ference with the terrain. This approach conrms the validity

of applying variant analysis as early as the conceptual stage.

The choice of the immediate vicinity of the gunpowder

laboratory as the area for variant analysis was dictated by

three methodological considerations. Firstly, a coherent,

high-resolution point cloud was available for this area. This

enabled reliable modelling of the terrain.

Secondly, the aim of the study was not to identify the tac-

tically optimal location for fortications. Such an aim would

have required an analysis of an area extending well beyond

Boyen Fortress. It would also have required consideration

of military doctrine and current defence service guidelines.

Instead, the aim was to demonstrate a methodology for ex-

ploring variants of site interventions within a BIM environ-

ment, using a specic, well-documented case study.

Thirdly, the analysis in the immediate vicinity of the his-

toric monument highlights a signicant conservation con-

straint. None of the proposed variants is recommended for

 BIMinassessingthesuitabilityofhistoricforticationsformilitarypurposes 85

in clear conict with the preserved historic earthworks of

Boyen’s Fortress. This applies to the cross-walls, parapets

and the structure of the moat. Carrying out such works in

the current condition of the site would lead to irreversible

damage to the historic fabric. Deeper earthworks could also

endanger the underground, masonry section of the labora-

tory. From a conservation perspective, none of the options

presented can be recommended for implementation. This

would require a prior comprehensive Heritage Impact As-

sessment (HIA). The consent of the relevant heritage con-

servation ocer would also be required.

It is precisely this observation that constitutes one of the

main conclusions of this study. Variant analysis within a BIM

implementation given the current condition of the site. This

is due to the risk of damage to the preserved earthworks,

such as cross-walls and parapets, as well as the under-

ground section of the laboratory. The analysis of alternative

options is therefore illustrative in nature. It demonstrates

how a BIM-based analysis of alternative options enables

the identication of conicts with the historic fabric even

before a decision is made on site.

Discussion

The proposed options for the location of the trenches

in the immediate vicinity of the gunpowder laboratory are

a b

Fig. 5. A simulation of a trench created using: a) Nano Banana Pro – variant 1 (source: Nano Banana Pro – Gemini 2.5 Flash Image; Google),

b) ChatGPT – variant 2 (source: ChatGPT; OpenAI)

Il. 5. Symulacja okopu utworzona za pomocą: a) Nano Banana Pro – wariant 1 (źródło: Nano Banana Pro – Gemini 2.5 Flash Image; Google),

b) ChatGPT – wariant 2 (źródło: ChatGPT; OpenAI)

Variant 1 (Fig. 5a) Variant 2 (Fig. 5b)

Location northern and eastern parts of the site northern part of the site

Length long short

Course very winding, zigzagging zigzagging

Primary function defensive and communication observation and combat

Defensive potential high medium

Extent of earthworks greater smaller

Design assumptions

a description incorporating a long, zigzagging trench

adapted to the natural terrain, providing cover for positions

and enabling communication along the line of defence

a description involving a short, zigzagging trench

with a simplified layout, designed for rapid construction

and to serve as an observation and cover post

Compliance with Ministry of

the Interior and Administration

Regulation 2025

prompt specifications in accordance with

the dimensions (height 150–220 cm, width 80–120 cm);

illustrative representation

the design of the prompt complies with the dimensions

(height 150–220 cm, width 80–120 cm);

illustrative representation

Table 1. Evaluation criteria and comparison of conceptual variants of trenches generated using AI tools (elaborated by K. Zelkowska, M. Żach)

Tabela 1. Kryteria oceny i porównanie koncepcyjnych wariantów okopów wygenerowanych z wykorzystaniem narzędzi AI

(oprac. K. Zelkowska, M. Żach)

86 Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

in the event of a building’s destruction as a result of mil-

itary action or other causes, a model created beforehand

can serve as digital documentation to support the recon-

struction or repair process. HBIM demonstrates just how

important it is to have models of historic build

ings. Tradi-

tional methods of documenting historic buildings, such as

technical drawings, textual descriptions or photographs, are

often incomplete, scattered and insucient for conducting

comprehensive spatial analyses and making conservation

decisions (Penjor et al. 2024). Furthermore, it is worth not-

ing that, in the case of reconstruc

tion, it is much easier to

use a model previously created using a point cloud whilst

the building still stands, rather

than using individual pho-

tographs, from which a model is only created after a great

deal of work. It should be noted, however, that the model

developed for the Boyen Fortress gunpowder laboratory is

simplied in nature and does not constitute a full HBIM

model. The data was obtained mainly from aerial views,

which limits the accuracy of the representation of the build-

ing’s architectural and structural details. To develop a mod-

el with a higher level of detail, it would be necessary to

carry out a scan from ground level and a detailed survey of

the site. At the same time, the level of detail adopted proved

sucient to achieve the aim of the study, which was to ex-

plore dierent options and visualise the potential location

of the trenches.

Signicant tool-related limitations were identied during

the implementation of the project activities. Autodesk Revit,

as a parametric environment, is not fully suited to modelling

and editing terrain with complex topography and a large

number of points. Attempts to adjust the terrain levelling

proved particularly problematic, placing a signicant com-

putational load on the programme and, con sequently, lead-

ing to instability and crashes. In practice, this made it di-

cult to make any corrections to the terrain model. A solution

in such a situation would be to use other tools dedicated to

working with terrain models, and to import simplied ge-

ometry into the BIM environment in order to improve the

quality of the work. However, in the work carried out in

this article, this approach was not adopted, which limited

the ability to model and modify visualisations of trench

variants. Consequently, a key element was the use of arti-

cial intelligence-based tools to visualise trench variants. The

use of AI enabled the rapid presentation of spatial concepts

without the need to manually model all variants, which sig-

nicantly reduced the time required for the work. These vi-

sualisations can be regarded as a stage preceding the actual

design process, allowing the selection of the most rational

solution for further, manual development within a BIM en-

vironment. It should be emphasised, however, that AI-gen-

erated visualisations are for illustrative purposes only and

do not constitute a professional simulation. Solutions based

on generative articial intelligence are therefore used in the

early stages of the design process, primarily at the concep-

tual stage, where they enable the rapid generation of new

ideas, thereby signicantly increasing the eciency of de-

sign work (Li et al. 2025). Nevertheless, the rapid devel-

opment of articial intelligence suggests that, in the future,

these tools may oer an increasingly high level of accuracy

and become a signicant aid in the design process.

environment allows for the early identication of conicts

with the historic fabric. The digital model reveals spatial

conicts which, at the preliminary concept stage, would not

be obvious without visualisation. In this way, BIM becomes

a tool for protecting heritage from hasty planning deci-

sions. It is not, however, a tool that facilitates intervention.

From the perspective of conservation practice, it would

be most appropriate to locate any contemporary fortica-

tions outside the boundaries of the historic complex. Vari-

ation analysis in BIM enables a rapid comparison of such

alternatives outside the fortress with options within it. This

provides conservation arguments in favour of the former

solution.

The use of the full BIM suite and AI tools to design

a single, limited section of earthworks could, in isolation,

be disproportionate to the scale of the problem. In this pa-

per, however, the scope of the analysis was determined by

its demonstrative function. The selected site and the limited

length of the variants allowed for a clear presentation of the

methodology within a single academic article.

The true value of the proposed approach only becomes

ap parent when applied on a larger scale. This applies to ana-

lyses covering entire fortication complexes, networks of

critical infrastructure, or cultural heritage portfolios. In such

cases, a unied methodology enables a systematic com-

parison of variants across multiple sites. In this sense, the

gunpowder laboratory at Boyen Fortress serves as a proof-

of-concept in this work, rather than the main focus of the

project investment.

BIM models of fortications, whether simplied or more

detailed, can serve as a valuable tool to support spatial anal-

yses, the evaluation of alternatives and the assessment of

potential scenarios. It is particularly advantageous to pre-

pare such models and analyses at a time when there is no

immediate threat and when it is possible to collect and anal-

yse data without time pressure. It is worth emphasising that

creating a simplied BIM model, based on a point cloud,

does not require highly advanced HBIM expertise. Nowa-

days, spatial data does not have to be acquired exclusively

using expensive platforms, and the equipment used for data

collection can often be signicantly cheaper (commercial

drones or even a mobile phone and a suitable app). This

lowers the technological barrier to entry and enables organ-

isations with limited technical resources to carry out such

work. It should be noted, however, that the equipment used

aects the accuracy and reliability of subsequent analyses,

which should be taken into account when interpreting the

results. Carrying out scans and creating models during pe-

riods when there is no threat brings many benets. The n-

ished model can be immediately used to carry out concep-

tual analyses, explore alternatives, and develop action plans

which, in the event of a conict, could be implemented

much more quickly and eciently. From a national security

perspective, an unquestionable advantage of using a BIM

model is precisely the ability to carry out scenario analyses,

compare options for eld interventions, assess their spa-

tial consequences, identify potential problems even before

eld operations commence, and conduct preliminary risk

predictions. BIM can thus serve as a tool to support de-

cision-making processes in crisis situations. Furthermore,

 BIMinassessingthesuitabilityofhistoricforticationsformilitarypurposes 87

Conclusions

The topic addressed in this article – the use of BIM to

assess and prepare historic fortications formerly used for

military purposes with a view to conducting variant analy-

ses of defence simulations – is rarely discussed in contem-

porary specialist literature or public debate, despite the fact

that it combines two important issues: heritage conserva-

tion and national security. Given the current geopolitical

situation, greater attention should be paid to the issues of

national security and preparedness, including the prepara-

tion of historic sites for potential defence in the event of

an armed conict. It is worth considering the creation of

digital models of sites during peacetime as a preventative

measure, which can be used in a crisis situation for rapid

scenario analysis and the evaluation of alternative concepts

without the need to commence survey work on a given site

under pressure and logistical constraints. In this context,

BIM serves as a useful tool for supporting variant planning

and spatial risk assessment, as it enables us to organise key

information regarding the geometry of a structure and its

surroundings. It would therefore be worth considering the

creation, in peacetime, of a database of BIM models of crit

ical infrastructure and historic buildings, which would be

regularly updated as part of heritage protection and crisis

management. Such a database would enhance the state’s

preparedness in the event of a conict and would also sup-

port conservation objectives. It should be noted that the

creation of such a database should be preceded by the de-

nition of clear criteria and standards, so that the models it

contains are as standardised as possible and ready for im-

mediate use without the need for corrections, which would

facilitate and accelerate work on the collected models.

Translated by

AndrzejSzymonBorkowski

References

Borkowski, Andrzej Szymon, and Alicja Kubrat. “Integration of Laser

Scanning, Digital Photogrammetry and BIM Technology: A Re-

view and Case Studies.” Eng 5, no. 4 (2024): 2395–409. https://doi.

org/10.3390/eng5040125.

Borkowski, Andrzej Szymon, and Patrycja Olszewska. “Model HBIM jako

cyfrowy bliźniak na przykładzie Chaty z Gawrych w Skansenie Kur-

piowskim w Nowogrodzie.” WiadomościKonserwatorskie –Journal

of Heritage Conservation 77 (2024): 137–48. https://doi.org/

10.48234/

WK77GAWRYCH.

Dore, Conor, and Maurice Murphy. “Current State of the Art Histor-

ic Building Information Modelling.” International Archives of the

Photogrammetry, RemoteSensingandSpatialInformationSciences

XLII-2/W5 (2017): 185–92. https://doi.org/10.5194/isprs-archives-

XLII-2-W5-185-2017.

Doyle, Peter. “Trench Construction and Engineering Geology on the West-

ern Front, 1914–18.” In MilitaryAspects ofGeology: Fortication,

Excavation and Terrain Evaluation, edited by Edward P.F. Rose, Judy

Ehlen and Ursula L. Lawrence. Geological Society of London, 2019.

https://doi.org/10.1144/SP473.6.

Franczak, Paweł, and Witold Jucha. “Odtworzenie przebiegu linii oko-

pów z II wojny światowej (OKH Stellung b1) w Paśmie Jałowiec-

kim i Grupie Mędralowej na podstawie numerycznego modelu terenu

z danych LiDAR i badań terenowych.” Acta Universitatis Lodzien-

sis. FoliaGeographica Socio-Oeconomica, no. 22 (2015): 87–108.

https://doi.org/10.18778/1508-1117.22.05.

Krahel, Sebastian. “Twierdza Boyen (Giżycko) – fortykacja jako atrakcja

turystyczna.” PraceiStudiaGeograczne 32 (2003): 99–115.

Li, Chengyuan, Tianyu Zhang, Xusheng Du, Ye Zhang, and Haoran Xie.

“Generative AI Models for Dierent Steps in Architectural Design:

A Literature Review.” FrontiersofArchitecturalResearch 14, no. 3

(2025): 759–83. https://doi.org/10.1016/j.foar.2024.10.001.

Liu, Xiaoyu, Xugang Lian, Wenfu Yang, Fan Wang, Yu Han, and Ya-

fei Zhang. “Accuracy Assessment of a UAV Direct Georeferencing

Method and Impact of the Conguration of Ground Control Points.”

Drones 6, no. 2 (2022): 30. https://doi.org/10.3390/drones6020030.

López, Facundo José, Pedro M. Lerones, José Llamas, Jaime Gómez-

García-Bermejo, and Eduardo Zalama. “A Review of Heritage

Build ing Information Modeling (H-BIM).” Multimodal Technolo-

gies and Interaction 2, no. 2 (2018): 21. https://doi.org/10.3390/

mti2020021.

Murphy, Maurice, Eugene McGovern, and Sara Pavia. “Historic Building

Information Modelling (HBIM).” Structural Survey 27, no. 4 (2009):

311–27. https://doi.org/10.1108/02630800910985108.

Penjor, Tshering, Saeed Banihashemi, Aso Hajirasouli, and Hamed Golzad.

“Heritage Building Information Modeling (HBIM) for Heritage Con-

servation: Framework of Challenges, Gaps, and Existing Limitations

of HBIM.” DigitalApplicationsinArchaeologyandCulturalHeritage

35 (2024): e00366. https://doi.org/10.1016/j.daach.2024.e00366.

“Rozporządzenie Ministra Spraw Wewnętrznych i Administracji z dnia 27

czerwca 2025 r. w sprawie budowli ochronnych i ukryć.” Dz.U. 2025

poz. 932. Published July 14, 2025. Accessed November 17, 2025, at

https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU20250000932.

Twierdza Boyen Giżycko. “14. Laboratorium Prochowe.” Published n.d.

Accessed November 7, 2025, at https://przewodniktwierdza.gizycko.

pl/lista-obiektow/14-laboratorium-prochowe.

“Twierdza Włocławek. Okopy – najpowszechniejsze umocnienia ‘twier-

dzy’.” Published 2025. Accessed November 17, 2025, at https://www.

twierdzawloclawek.pl/elementor-246/.

Streszczenie

BIM w ocenie przydatności zabytkowych fortykacji do celów militarnych

na przykładzie laboratorium prochowego Twierdzy Boyen w Giżycku

Building Information Modeling (BIM) znajduje coraz szersze zastosowanie w dokumentowaniu, analizie oraz ochronie obiektów zabytkowych,

umożliwiając tworzenie cyfrowych modeli, które następnie mogą być wykorzystywane jako podstawa do różnorodnych analiz przestrzennych. W kon-

tekście obiektów fortykacyjnych szczególne znaczenie ma możliwość wykorzystania takich modeli do oceny ich potencjalnej przydatności w scenariu-

szach obronnych. W artykule na przykładzie laboratorium prochowego Twierdzy Boyen w Giżycku (Polska) przedstawiono zastosowanie uproszczone-

go modelu BIM jako narzędzia wspierającego analizę powiązań obiektu z otaczającym terenem oraz koncepcyjne wariantowanie ingerencji terenowych,

88 Andrzej Szymon Borkowski, Klaudia Zelkowska, Małgorzata Żach

takich jak okopy. Cyfrowy model obiektu i jego otoczenia, uzupełniony o analizy przestrzenne, umożliwił ocenę uwarunkowań przestrzennych, identy-

kację potencjalnych kierunków zagrożeń oraz porównanie możliwych rozwiązań lokalizacji i wyglądu umocnień ziemnych. Ze względu na ograniczenia

środowiska BIM w zakresie modelowania złożonej geometrii terenu na etapie koncepcyjnym wykorzystano narzędzia generatywnej sztucznej inteli-

gencji do wizualizacji wariantów potencjalnych umocnień ziemnych. Uzyskane rezultaty wskazują, że nawet uproszczone modele BIM mogą stanowić

użyteczną podstawę do analiz wariantowych oraz dokumentacji obiektów zabytkowych. Jednocześnie narzędzia oparte na sztucznej inteligencji znajdują

zastosowanie przede wszystkim na wczesnych etapach prac koncepcyjnych, wspierając proces podejmowania decyzji poprzez szybkie przedstawienie

możliwych rozwiązań, przy zachowaniu ich poglądowego charakteru.

Słowa kluczowe: BIM, Twierdza Boyen, chmura punktów, okop, sztuczna inteligencja, laboratorium prochowe