Case Studies in Thermal Engineering 60 (2024) 104601

Contents lists available at ScienceDirect

Case Studies in Thermal Engineering

journal homepage: www.elsevier.com/locate/csite

Process Integration and Electrification for retrofit: Case studies of

milk evaporator systems
Benjamin James Lincoln a, Lana Kong a, Florian Schlosser b,
Timothy Gordon Walmsley a, *
a Ahuora – Centre for Smart Energy Systems, School of Engineering, University of Waikato, Hamilton, 3240, New Zealand
b Department of Energy System Technologies, Paderborn University, Warburger Str. 100, 33098, Paderborn, Germany

ARTICLE INFO ABSTRACT

Handling Editor: Huihe Qiu Industrial Process Integration and Electrification represents a critical path to harnessing renew-
able electricity for a net-zero future, particularly through advancements in heat pump and vapour
Keywords:
recompression technologies for industrial applications. This paper addresses the challenge of de-
Process integration
carbonising an energy-intensive aspect of dairy processing plants, milk evaporation, which is a
Process retrofit
Bridge analysis
critical processing step in milk powder production. Employing three different milk evaporator
Process electrification case studies, it introduces a retrofit methodology that merges Process Integration with Electrifica-
Heat pumps tion technologies to achieve energy, emissions and cost savings. The study leverages Heat Pump
Dairy industry Bridge Analysis, extended to evaluate process unit operations, and work consumption and gener-
ation, facilitating a comprehensive retrofit analysis for optimal energy use. The study concludes
with several retrofit solutions, notably replacing steam ejectors with mechanical vapour recom-
pression fans, resulting in energy savings of 13.7 %–41.6 % and emissions reductions of
14.5 %–47.3 %. This approach offers a holistic framework for industrial electrification, demon-
strating significant efficiency and sustainability gains in dairy processing.

Nomenclature
Abbreviations
COND Condensate
COP Coefficient of Performance
COW Condensate of Whey water
DSI Direct Steam Injection
EGCC Exchanger Grand Composite Curve
ESCC Exchanger Shifted Composite Curve
ETD Energy Transfer Diagram
HEN Heat Exchanger Networks
HSDT Heat Surplus Deficit Table
IPS Intermediate Pressure Steam
LCOH Levelised Cost of Heating
(continued on next page)

* Corresponding author.
E-mail address: tim.walmsley@waikato.ac.nz (T.G. Walmsley).

https://doi.org/10.1016/j.csite.2024.104601
Received 16 February 2024; Received in revised form 12 May 2024; Accepted 25 May 2024
Available online 26 May 2024
2214-157X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

(continued)
Lp Low Pressure (Steam)
METD Modified Energy Transfer Diagram
MPS Medium Pressure Steam
MVR Mechanical Vapour Recompression
PA Pinch Analysis
PE Process Electrification
PI Process Integration
PI&E Process Integration and Electrification
TVR Thermal Vapour Recompression
Vap Vapour
Symbols
ΔTcont Contribution to the approach temperature (°C)
a Annuity factor, fitting parameter
cI Investment costs
E Effect
f Factor
H Duty (kW)
i Run indices
N Depreciation period
p Energy carrier price $/kWh
Pcom Power consumption (kW)
Q Cumulative enthalpy of the steam (kJ)
Q̇ Enthalpy flow of the steam (kW)
rp Electricity-to-reference price ratio
t Run time (s)
Subscripts and superscripts
* Shifted
cap Capital-related
cooling Cooling
dem Demand-related
el Electricity
Evap Evaporator
FL Full load
M Maintenance
ref Reference
op Operation-related
s Supply
sav Savings
t Target
water evap Water evaporated

1. Introduction
The need for a rapid global transition towards renewable energy continues to build pressure on the industrial sector to develop
large-scale, low emissions solutions. The industrial sector makes up 37 % of the total energy-related emissions globally [1], which is a
direct result of the sector's heavy reliance on fossil fuels to generate process heat.
In New Zealand, for example, the dairy industry alone contributes to 14 % of the total process heat demand, which is largely sup-
plied by coal and natural gas [2]. Much of this process heat is consumed during the evaporation and spray drying processes. During
evaporation, pasteurised milk is fed through the tubes of a falling film evaporator and is heated using fossil steam to evaporate the
water from the milk. The concentrated milk is then fed to a second evaporator, or effect, where the excess vapour from the first effect
is used to heat the second effect to further concentrate the milk, reducing the amount of fossil steam required to reach the desired con-
centration.
In the 1980s, evaporator plants typically cascaded vapour through four or seven effects to increase the efficiency of the system.
Since then, significant improvements in energy and spatial efficiency for evaporators have been achieved with the development of
vapour recompression technologies. The earlier multi-effect evaporator systems generally had low levels of integration which re-
sulted in substantial low-grade waste heat rejection (560 MJ/twater evap). The Coefficient of Performance (COP) or ratio between the
evaporation energy required and the energy that was supplied by fossil steam for a four-effect cascade would be approximately 4 [3].
In the early 1990s, Thermal Vapour Recompression (TVR) was introduced that gave the ability to recover low-grade waste heat using
a relatively small amount of high-grade heat, which subsequently raised the COP to 7 and reduced the required number of effects re-
quired to satisfy the process requirements. Mechanical Vapour Recompression (MVR) integrated with TVR technology became the
most common design by the mid-1990s [3]. With MVR, the low-grade waste heat could be recovered by recompressing it through a
blower, compressor, or fan, at very high efficiencies using only electrical work as an input. The MVR/TVR systems usually comprised
of either two or three effect arrangements. These systems would be well-integrated such that little to no heat rejection occurred, re-
sulting in much higher COPs of approximately 30. Additional energy efficiency improvements have been marginal in recent years,
causing the two and three effect MVR/TVR evaporators to be the most widely used evaporator design.

2
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

While low-emissions technologies are becoming more prevalent, emissions reduction for the industrial sector still presents a com-
plex challenge in developing solutions that balance the environmental and economic needs of the sector. As greenfield sites become
far less common, these solutions must be flexible enough for the retrofit of existing sites. Multiple studies have investigated general
evaporation retrofits. Boldyryev et al. [4] utilises a Pinch Analysis (PA) approach to reduce primary energy use and emission for a cal-
cium chloride evaporator. Their method consisted of first simulating the existing process to establish a base case, then via the use of
conventional PA methods, a minimum energy target and savings opportunities were determined. The authors then proceed to test
said opportunities via updated simulation models to test efficacy. The paper shows how process simulation and process integration
can work in parallel to provide solid and quantified retrofit options, however, the paper was limited by conventional PA techniques as
it was more difficult to determine where inefficiencies were.
Another study investigated retrofit opportunities for the downstream processing of amino acids via the application of Process Inte-
gration (PI) methods [5]. Similar to the previous study the paper combines process simulation with PI to evaluate technical feasibility
iteratively. However, the paper determined that conventional PI retrofit pathways such as additional heat exchangers or repiping
would provide no practical savings for the current system as the system met theoretical targets. Instead, Park et al. focuses on process
modifications, proposing the introduction of a TVR unit, rerouting steam mixing, and feed stream redistribution. These savings were
found through good engineering judgement and simulation; however, the lack of a general method and heat pump integration was a
key limitation.
The most recent study into milk evaporator efficiency was conducted by Tsochatzidi et al. [6] where the authors used model based
optimisation to reduce energy consumption for multiple evaporator arrangements. The paper found that electricity and capital costs
proved to be a dominating factor when deciding between TVR and MVR technology. The paper however did not consider retrofit
cases as its aim was to compare minimisation strategies.
Another common theme across the current evaporator studies was a focus on energy minimisation rather than efficient electrifica-
tion for decarbonisation.
Process Integration (PI) techniques such as Pinch Analysis (PA) [7] have proven to be an effective method for reducing energy de-
mand from a process. In particular, the Grand Composite Curve is an effective tool to identify areas of high-grade heat deficit, and
low-grade heat surplus that inform areas for the appropriate placement of vapour recompression [3]. Process Electrification (PE) is
gaining popularity as a method to displace fossil fuel usage and describes the method for developing full or partially electrified
process designs. PE includes a range of technological solutions, such as heat pumps, electrode boilers, and electricity-driven
processes, that enable fuel switching from fossil fuels to electricity. PE is particularly effective in locations where the electricity sup-
ply is largely renewable, such as New Zealand which has 84 % renewable electricity generation [8]. It has also been found to have sig-
nificant strategic value for low-temperature heat industries, e.g., dairy processing [9] and food processing [10]. Full PE is now eco-
nomically viable for many low-temperature processes using existing technologies when integrated appropriately. However, the effi-
cient and economical integration of electrification technology requires systematic analysis to ensure appropriate placement, opera-
tion, and design.
Process Integration and Electrification (PI&E) aims to harness Process Integration methodologies to maximise system efficiency in
transitioning systems to rely more on electricity as the primary energy source instead of fossil fuels. Lincoln et al. [11] presented an it-
erative approach to PI&E that combines many existing PI tools and concepts, together with Process Simulation, to synthesise a system-
atic greenfield design method. The PI&E method generated efficient designs that meet the targets for Exergy Pinch Analysis. The use
of Exergy Pinch Analysis can lead to lower bound work targets by considering the opportunity for partially cutting heat pockets (heat
recovery) to maximise the integration of heat pumps.
Kim et al. [12] also identifies PI as the most appropriate strategy to integrate electrified sources of energy. The authors focused on
high temperature systems where steam and hot oil utilities were the suitable technology. Although the paper found three heuristics to
minimise cost and risk, the authors found it challenging to compete with traditional generation without a reduction in electricity
prices. Lower temperature processes can mitigate this issue via high COP heat pumps.
Cui et al. [13] presents a successful PE case study highlighting how electric technologies that leverage self-heat recuperation (or
heat upgrading) combined with PI techniques can provide significant cost and operational advantage over thermally driven systems.
This work has added to Xia et al. who all shows the benefits of self-recuperation in pressure swing distillation.
The next challenge, addressed in this paper, is to extend the PI&E concept to the retrofit of existing processes and Heat Exchanger
Networks (HEN) with both increased efficiency and degree of electrification.
This paper aims to develop a comprehensive PI&E retrofit method that includes potential modification of the process (e.g., operat-
ing set-point, change in technology), HEN (e.g., heat transfer enhancement), and process electrification (e.g., MVR). The method rep-
resents an extension of Bridge Analysis [14], heat pump Bridge Analysis and the Modified Energy Transfer Diagram (METD) [15]. The
case studies for this work include three typical milk evaporator designs that are found in the New Zealand dairy industry.
The novel contributions of this work to the literature include:
• The development of an improved PI&E retrofit method using an extension of heat pump bridge analysis to include the process
units. The inclusion of process unit heat flows allows for process modifications to be considered. A key aspect identified in
previous literature to electrify processes efficiently.
• The demonstration of applying the method to multiple related case studies
• The design of common retrofit solutions for milk evaporator plants to transition to renewable electricity fuel.

3
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

2. Overview of Modified Bridge Analysis tools

Bridge Analysis and the Energy Transfer Diagram (ETD) was first introduced by Bonhivers et al. [14]. Walmsley et al. [15] pre-
sented a Modified Bridge Analysis and the Modified Energy Transfer Diagram (METD). The METD was comprised of “stacked” Ex-
changer Grand Composite Curves (EGCCs) for individual exchangers and aimed to represent the net heat transfer within an existing
HEN. Unlike the ETDs presented originally by Bonhivers et al. [14], stacking the EGCCs to form the METD enabled the diagram to
maintain the structure of a HEN, which is advantageous for providing insight into potential inefficiencies within the HEN. Lal et al.
[16] used the concepts of Modified Bridge Analysis to develop the Heat Surplus Deficit Table (HSDT), which improves the identifica-
tion and quantification of retrofit bridges (Fig. 1). It was demonstrated across three case studies that the METD was simpler to under-
stand than the ETD due to its tabular presentation. However, with a problem scale of more than 10 streams and an existing HEN, the
number of possible retrofit options grow exponentially, which makes the use of either graphical or tabular representations very te-
dious and impractical for identifying and analysing specific opportunities.
Walmsley et al. [17] outlines how areas of heat surplus can be easily identified and transferred to areas of heat deficit at a lower
temperature interval. The energy surpluses (or coolers) can be paired to a recovery exchanger or directly to a heater (area of heat
deficit) with the maximum feasible heat transfer recorded for each pairing as highlighted in Fig. 2. The minimum value of the feasible

Fig. 1. The relationship between the Exchanger Shifted (a) Composite Curve (ESCC), (b) EGCC, and (c) E-PTA for recovery exchanger as shown in Walmsley et al. [15].

Fig. 2. HSDT Bridge identification illustration from Lal et al. [16].

4
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

heat transfers will determine the maximum achievable energy savings of the retrofit bridge. Schlosser et al. [18] adapted the HSDT to
include considerations for heat pump retrofit. Similar to Lal et al. [16], the method presented by Schlosser et al. [18] connects higher
temperature surpluses to lower temperature deficits, however, the addition of heat pumps to the method allow for the lower tempera-
ture surplus to be integrated with higher temperature deficits at the cost of work. The work is estimated using Eq. (1), where COP can
be estimated using COP correlations, such as those presented by Schlosser et al. [19], or isentropic efficiencies for closed or open cycle
heat pumps respectively.

HEvap
W= (Eq 1)
COP − 1

The method is modified again in this paper to account for Process Electrification, where the objective is the economic replacement
of fossil fuel generated steam, rather than a primary focus on utility savings.

3. Method
The purpose of this study is to produce an intuitive PI&E retrofit method that will be used to determine general cost-effective elec-
trification retrofit strategies for the milk evaporation process. Fig. 3 presents a flowchart of the method with descriptions of the key
steps described below.
1. Gather process design data for the existing process.
2. Define PE targets for full or partial electrification. The targets could also be defined as a percentage of steam supplied by fossil
fuels, referred to as fossil steam, to be replaced. The targets should consider practical aspects of the process, such as scheduling,

Fig. 3. Process Integration and Electrification retrofit design method through incremental changes.

5
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

process requirements and physical dimensions of the equipment. The targets should also consider existing electric technologies
and operational costs.
3. Simulate the design for the existing process with a process simulator to generate a reliable and complete data set.
4. Formulate the HSDT that includes both the existing process HEN and internal process streams and identify areas of heat recovery
potential.
5. Evaluate the waste heat streams to determine if a retrofit bridge could “unlock” heat at a higher temperature interval. If so,
follow conventional Bridge Analysis to form the partial bridge. If not, continue to the next step.
6. Upgrade waste heat to the required temperature to replace steam from fossil fuel boilers. This step requires the selection of
appropriate heat pump technology and an estimation of the required work using COP correlations.
7. Determine the operating cost impact (i.e., fossil fuel replacement with electricity) of the proposed retrofit solution. Calculate
levelised cost of heating (LCOH) summarising capital- (Acap), demand- (Adem) and operation-related (Aop) costs based on Eq. (2),
Eq. (3) and Eq.4 according to the steam savings ( Q̇sav ) and the MVR compressor capacity Pcom from Table 2 and economic
parameters from Table 3 [20]. The capital costs were calculated from the annuity factor (a) , investment costs (cI) for the run index
(i) and the power consumption (Pcom) where the annuity factor (a) is a fitting parameter calculated as shown in Eq. (3). The
demand related costs were calculated from the reference energy carrier price (pref), the electricity-to-reference price ratio (rp) and
the run time at full load (tFL). The operational costs were calculated from a maintenance factor (fM) for the run index, the
investment cost and the power consumption.

Acap + Adem + Aop a ⋅ fc,Inst ⋅ cI,i ⋅ Pcom + pref ⋅ rp ⋅ tFL ⋅ Pcom + fM,i ⋅ cI,i ⋅ Pcom
LCOHi = = (2)
Qsav Q sav ⋅ tFL

a = (1 + i)N ∙ i ∕ (1 + i)N − 1 (3)

( ) ( )

8. Modify the simulation with the new retrofit solution and determine if the new design meets the desired PE level. Iterate through
the method to determine additional retrofit improvements until the desire PE target has been reached.
The HSDT algorithm and method were implemented in the OpenPinch Excel Workbook [21]. Readers can contact the correspond-
ing author for access and more information. The process simulation was undertaken using Aspen HYSYS [22].

4. Case studies
This study analyses three typical milk evaporator configurations to demonstrate the method and highlight common cost-effective
solutions across the different designs. Currently, two predominant evaporation process designs are used in modern dairy plants. The
first design (Fig. 4) uses one MVR effect and one TVR effect. The second design (Fig. 5) uses one MVR effect followed by two effects
with TVR. A four-effect TVR cascade design (Fig. 6) is also included as a case study to represent an older evaporator design that was
typical for the 1990's and are still operational today in some dairy factories. Designs with a greater number of effects are less common
as it is preferable to minimise capital cost and area, however, the method can also be applied to designs with a greater number of ef-
fects if required. Data for each of the configurations are based off mass and energy balances using Aspen HYSYS that apply consistent
boundary conditions (where possible) and processing parameters. Processing parameters were derived from prior research studies
[3].
The boundary conditions for the evaporation system are the same across the three designs and define the temperatures, composi-
tions, and mass flowrates at the inlet to the preheating section, the inlet to the first effect and the outlet of the evaporation system. The
inlet conditions to the preheating section are 247 t/h of whole milk with 14.5 wt% solids entering at a temperature of 8 °C. The milk

Fig. 4. Case 1 – two-effect MVR-TVR effect evaporation system.

6
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

Fig. 5. Case 2 – three effect MVR-TVR evaporation system.

Fig. 6. Case 3 – four effect TVR evaporation system.

is preheated to 95 °C with the final heating achieved through direct steam injection for rapid heating and held for about 30 s for heat
treatment, before it enters the first evaporator effect. The final concentrate conditions exiting the evaporator are 51 wt% solids and at
a temperature of 79 °C.
In Case 1 (Fig. 4) and Case 2 (Fig. 5), a flash vessel for heat recovery is included between the heat treatment process and the first
effect. The flash process recovers 3152 kW. Across the first effect a mean temperature driving force of 4 °C is maintained for both
Cases 1 and 2 using an MVR fan, whereas the final TVR effects requires a driving force of 7 °C due the increase in viscosity of the milk.
Case 2 has an additional effect that receives upgraded vapour from the TVR on the final effect. The additional effect then cascades
vapour to the final effect. Case 3 (Fig. 6) has four cascade effects where vapour exiting the final effect is upgraded using a steam ejec-
tor, or TVR unit, and medium-pressure steam to supply the first effect. All effects operate at a mean 4 °C temperature lift except the
last which operates at a 7 °C lift.
In this study, the evaporation plant is considered in isolation from the rest of the plant, however, in future research, the considera-
tion of the entire plant when integrating may lead to further electrification strategies. The data for the case studies was generated
through process simulation, previously published articles, and industrial experience of the authors in analysing dairy processes.

5. Results and discussion

The three designs were simulated using Aspen HYSYS. The data generated through simulation is then validated against the process
design data to determine whether the simulation is representative of the existing process. This is necessary to ensure that retrofit solu-
tions developed in simulation are applicable to the real process. The validated data is then processed using the OpenPinch Excel
Workbook to generate the HSDT and undertake the retrofit analysis. To demonstrate the method, the results for Case 1 are presented
in a detailed step-by-step manner. Throughout each step, a discussion of the method and the subsequent retrofit design concepts are
presented. The same method was also applied to Cases 2 and 3.

5.1. Simulation and expansion of the HSDT method

Conventionally, HSDTs typically focus on the HEN and is used to search for heat exchanger retrofit opportunities. However, ex-
panding the analysis to include process operations creates the opportunity to identify further emissions reduction and PE potential.
This paper presents a modification to stream data collection to allow for these internal balances of process operations. First, a system
boundary is drawn around the process to be included (e.g., Fig. 7). Unlike a heat exchanger, process operations usually have reac-
tions, separator and/or mixing of streams occurring within the unit. As a result, traditional techniques for representing stream data as
enthalpy differences between supply and target conditions cannot be applied. Likewise, the concept of a minimum approach tempera-
ture also cannot apply since mixing occurs. Instead, a reference temperature must be defined for the system, which is used as an
(imaginary) intermediate state, where inlet streams are “cooled” to the reference state (target) and outlet streams are “heated” from

7
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

Fig. 7. Example system boundary analysis of a process.

the reference state (supply). This concept can be applied to a heat exchanger to achieve the same result as conventional analysis. By
bringing all streams to a common reference temperature, process units can be mapped on a temperature-enthalpy plot as enclosed
pockets, similar to a heat exchanger. An enclosed pocket indicates that the heat surpluses and deficits around the operation is bal-
anced. An example set of stream data for a milk evaporator is presented in Table 1.

5.2. Identify waste heat and application of conventional bridge analysis

The resulting stream data is processed using the method proposed by Walmsley et al. [15] and converted into the HSDT format.
The paper presents a new format style for the table in Fig. 8, adding rows for utility usage and heat pump analysis. The new format ex-
plicitly quantifies the degree of PE on the diagram, allowing for quick appraisal of potential retrofit solutions.
A potential heat pump bridge is formed by integrating a portion of underutilised excess heat through a network of recovery ex-
changers and heat pump devices to replace fossil steam. Areas of underutilised heat (or areas of recovery potential) are identified
through one of two characteristics. 1) The heat surplus is connected to a cooling utility or 2) the heat surplus is utilised in a high-
emitting process operation (such as a TVR). Fig. 8 identified two areas of underutilised heat, the first being the 4118 kW of surplus
cooled in the COND Lp stream. The second is the 2,494 kW of heat surplus upgraded in the TVR E2 stream.
A key step in PE Retrofit is the minimisation of the temperature lift performed by the heat pump. A smaller temperature lift corre-
lates to a higher efficiency/COPs, minimising the required electricity consumption. Using conventional Bridge Analysis [16], a utility
path was created from areas of underutilised heat, through a set of recovery exchangers, to a heat surplus at a higher temperature in-
terval. The paper used a minimum approach temperature of 3 °C, however, due to the HSDTs having non-shifted temperatures, the
minimum approach temperature must be enforced by the user. The first match in Fig. 8 shows that only 2,335 kW out of the 3,132 kW
is at an approach temperature of 3 °C. Fig. 8 also shows a complete 2,355 kW heat bridge from the 57 °C COND Lp stream to the 72 °C
Vap Bleed heat surplus.

5.3. Integration of heat pump technology

It is important during the integration of heat pumps to consider the process requirements and conditions, such as direct or indirect
heating. These practical considerations will inform the type of heat pump technology best suited for the proposed retrofit solution.
For the example above, integrating a vapour stream to replace a Direct Steam Injection (DSI) is a form of direct heating which is suit-
able for the use of MVR. MVR provides the additional benefit of no intermediate refrigerant and, therefore, higher efficiencies can be
achieved. To estimate the electricity cost of a potential bridge, the heat pump COP was calculated. This can be done through heat
pump COP correlations [19] or isentropic efficiencies of MVRs. For the example problem, an isentropic efficiency of 85 % was used
based on available MVR technology [23]. The resulting COP for the 72 °C–98 °C temperature lift was calculated to be 10. It will cost
262 kW to upgrade the 2,355 kW of heat unlocked by the first retrofit bridge.

Table 1
Example stream data for the system boundary analysis of a process.

Stream Ts (°C) Tt (°C) ΔH (kW) ΔTcont (°C)

Effect 2 Entry 57 5 4,267.2 0

Milk Conc 2 5 57 2,806.6 0
TVR IPS 61.9 61.8 6,310.9 0
TVR IPS Sub Cooled 61.8 5 616.4 0
Water Vap 56.9 57 7,125.2 0
Water Vap Sub Cooled 5 56.9 644.5 0
COW Condensate 5 61.8 618.1 0

8
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

Fig. 8. Example Heat Pump Bridge Analysis for the two effect MVR-TVR Case.

To demonstrate the impact of smart heat pump integration in economic design, another retrofit option was considered. If the
retrofit choice was to instead upgrade the excess vapour going to the condenser (58 °C) up to the DSI temperature of 98 °C, the COP
will decrease substantially to a COP of approximately 3 when compared to upgrading vapour from 73 °C (COP is approximately 10).

5.4. Estimate the impact and iterate

Once the heat pump retrofit bridge was identified and the electrical cost was determined, the fossil steam was replaced with up-
graded vapour using the HSDT. The heat recovery potential and the resulting retrofit bridge was then removed from the diagram.
Subsequently, a new HSDT of the remaining heat available for retrofit was formed, which also showed the degree of electrification for
the new design. If the degree of electrification, the electrical cost and the number of retrofitted equipment required (each HSDT
stream that is matched requires a new heat exchanger or heat pump to be retrofitted) was satisfactory when compared to the PE target
set initially, no further iterations were required. The potential solution was then remodelled to determine its process viability. How-
ever, if the target was not satisfied then the resulting HSDT was used to determine a subsequent bridge match, furthering the electrifi-
cation of the process. This is demonstrated in Fig. 8.
For the example problem, the potential solution would consist of three new recovery exchangers and three MVRs in series to up-
grade the vapour. The first two heat exchangers are integrated by splitting the excess vapour, one to heat the raw milk from 8 °C to
12 °C then another to heat the milk from 45 °C to 55 °C. The third exchanger purchased will use the COW water to heat the milk from
55 °C to 60 °C. Then, all 3,758 kW of high temperature excess vapour bleed from effect 1 can then be upgraded to replace fossil steam
in the DSI.
As a summary, the steps taken to iterate through designs are as follows:
1 Produce the HSDT diagram.
2 Determine areas of under-utilised heat, for the example the cooling exchanger COND Lp has heat available at temperature
(57 °C) that it can be exchanged with the sink of recovery exchanger Vap Bleed (45 °C–54 °C).
3 COW preheat also has a heat sink at a low enough temperature (8 °C–30 °C) to accept heat from COND Lp unlocking additional
higher quality heat sources.
4 Utilise the unlocked heat to bridge to another lower temperature sink or upgrade the unlocked heat via heat pump technology.
For the above example, COW preheats source is not a vapour (MVR cannot be used) therefore, a heat bridge is created to the Vap
bleed sink, further unlocking high quality heat.
5 The unlocked 72 °C vapour is upgraded via MVR to replace the MP source used for DSI preheat.
6 Performance and costs are estimated.

5.5. General retrofit solutions

The results of applying the proposed retrofit methodology to the three case studies are shown in Table 2. Two distinctive retrofit
pathways were common across all three cases with an example process flow diagram for each proposed solution shown in Figs. 9 and

9
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

Table 2
Steam savings and electricity cost across the three case studies.

Design Retrofit Method Steam Savings Q̇sav Additional Electricity Pcom Energy Savings Emissions Savings

Two Effect MVR-TVR DSI Replacement 4467 kW 447 kW 38.2 % 44.6 %

Two Effect MVR-TVR TVR Replacement 4154 kW 126 kW 38.3 % 42.9 %
Three Effect MVR-TVR DSI Replacement 4175 kW 418 kW 37.0 % 43.2 %
Three Effect MVR-TVR TVR Replacement 3453 kW 115 kW 32.9 % 36.9 %
Four Effect TVR DSI Replacement 3109 kW 311 kW 13.7 % 14.5 %
Four Effect TVR TVR Replacement 10722 kW 2212 kW 41.6 % 47.3 %
DSI replacement ΔTlift ∼23 °C DSI replacement COP ∼10
TVR replacement ΔTlift ∼5 °C TVR replacement COP ∼50

10. The first solution involves DSI replacement as shown in the example above Fig. 8. The opportunity exists to take excess vapour, in-
tegrate recovery heat exchangers to unlock a higher temperature stream, then subsequently upgrade this vapour using MVRs for all
three cases. In all cases, the DSI replacement with MVR units achieved a relatively low work consumption compared to a DSI replace-
ment solution with an air sourced heat pump (ranging between 311 kW and 447 kW) whilst saving a large proportion of fossil steam
(between 3,109 kW and 4,467 kW).
The other viable retrofit strategy identified was TVR replacement. All cases presented involved heat surplus that was utilised in a
TVR which relies on fossil steam (high-emitting process unit operation) for upgrading low-pressure vapour to satisfy the required
heating for the effect. In the proposed retrofit strategy, the TVR can be replaced with an MVR, replacing the fossil steam with electric-
ity. The efficiency of an MVR (COP of 50 assuming an isentropic efficiency of 86 %) is much larger than the efficiency of the TVR
(equivalent to a COP of approximately 1.5). Therefore, less utility is required per unit of vapour to be upgraded. However, to achieve
the required heating for the effect, a larger mass flowrate of low-pressure vapour is required to compensate for the lower utility and
maintain the energy balance. Two out of three cases showed similar levels of steam savings for a significantly smaller electrical cost
and number of modifications compared to the DSI replacement solution. The benefit was increased substantially for the four effect
TVR case due to the reliance on fossil steam and TVR in older designs.
The retrofit strategies can also be used in conjunction with one another. The HSDT shows that the TVR replacement designs also
have remaining heat surpluses available from the COND Lp stream ranging between 500 kW for the two effect MVR-TVR, 1,000 kW

Fig. 9. Example of DSI replacement for the two effect MVR-TVR case.

Fig. 10. Example of TVR replacement solution for the two effect MVR-TVR case.

10
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

Table 3
Economic parameters for the LCOH calculation.

Parameter MVR Reference boiler

Investment costs, cI 1,633 NZD/kWel (for 100–1000 kW) 237.65 NZD/kWth [20]
792 NZD/kWel (for 1300–3000 kW)
Capital cost factor, fC,Inst 2 to 4
Reference energy carrier price, pref pel = rp ⋅ pref pref = 55.39
NZD/MWh
Full-load hours, tFL 6000 h/a (three-shift operation)
Interest factor, i 0.12
Depreciation period, N 20 a
Factor maintenance costs, fM,j 2.5 3.0

Fig. 11. LCOH of the three case studies for the TVR and DSI replacement design considering capital cost factors for planning, installation and infrastructure between
fC,Inst = 2 (solid lines) and fC,Inst = 4 (dashed lines).

for the three effect MVR-TVR, and 8,000 kW for the four-effect design. These respective heat surpluses are sufficient to partially or
fully replace the DSI, using the DSI replacement solution, effectively utilising both proposed viable retrofit strategies simultaneously.

5.6. Economic assessment

The economic assessment is based on the calculation of LCOH for different electricity-to-reference price ratios rp. The economic
parameters were assumed as shown Table 3. Capital costs for the MVRs are differentiated by mean values for different size classes and
depending on the specific electric capacity of the MVRs. For the planning, installation and infrastructure measure necessary of the
heat pump implementation, capital costs between 200 and 400 % of the investment costs were considered by the capital cost factor,
fC,Inst. Since the reference boiler is assumed as already existing and the retrofit of the heat pump is purely an efficiency measure, there
are no capital costs for the reference case. The investment costs are only used to calculate the operating costs of the reference system.
All other parameters were assumed according to Schlosser et al. [20].
Fig. 11 shows that the DSI replacement variants differ only slightly in terms of nominal electrical compressor capacity and COP,
which is why the LCOH are almost the same. The TVR replacement variants, by contrast, deviate from this. While the four effect vari-
ant is the most expensive over the majority of all price ratios, this variant even exceeds the fossil reference variant for high capital
costs (fC,Inst = 4) and high price ratios. However, the other two cases (two and three effect MVR-TVR design) show cost advantages.
In addition, they are less sensitive to the price ratio and capital costs due to their high efficiency and lower electrical compressor in-
put.

5.7. Other applications and limitations

Having a method that incorporates the processes heat cascade allows heat pump bridge analysis to be used in wider applications
where self-recuperation is available such as other chemical industries. The method is also assists in the integration of direct process
electrification technologies, for example, a pulsed electric field into a potato blancher as the internal heat flows are known.
Currently, the method only considers local process zones or individual process retrofits, causing a limitation when needing to elec-
trify a wider site with multiple zones. A combined method between bridge analysis and total site should be considered for future
work.

11
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

6. Conclusions
As more pressure is placed on industries such as dairy to reduce emissions and the interest in process electrification grows, intu-
itive and easy to use retrofit methodologies and solutions will become a necessity. This study focused on refining modified Bridge
Analysis into a Process Electrification Retrofit Methodology by including process unit operations and work consumption to reduce
emissions in a cost-effective manner. To demonstrate the method, investigations were performed on three milk evaporator case stud-
ies, which led to the proposal of two general retrofit solutions. These solutions were applicable and cost effective across all cases with
a TVR replacement being the simplest and most cost-effective solution overall, providing energy and emissions savings between 33.9
% - 47.6 % and 36.9 %–47.3 % respectively as well as significant cost advantages across a wide range of electricity prices. The case
studies showed that for effective Process Electrification, smart Process Integration and correct technology selection is key. In all cases,
the proposed replacement solutions were shown to be more cost effective than the reference system for electricity-to-reference price
ratios up to 4.2. The case study also exhibited the requirement to consider all elements of a plant; including process unit operations in
the retrofit study highlighted and quantified the most effective retrofit solution.
Overall, the modified Heat Pump Bridge methodology assisted in the discovery and quantification of retrofit solutions by allowing
appropriate heat exchange and heat pump placement to be easily highlighted.

CRediT authorship contribution statement

Benjamin James Lincoln: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualiza-
tion. Lana Kong: Writing – original draft, Visualization. Florian Schlosser: Writing – original draft, Visualization, Investigation,
Formal analysis. Timothy Gordon Walmsley: Writing – review & editing, Supervision, Software, Funding acquisition, Data cura-
tion, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability
Data will be made available on request.

Acknowledgements
This research has been supported by the programme “Ahuora: Delivering sustainable industry through smart process heat decar-
bonisation”, an Advanced Energy Technology Platform, funded by the New Zealand Ministry of Business, Innovation and Employ-
ment.

References
[1] IEA Key World Energy Statistics 2021 – Analysis Available online: https://www.iea.org/reports/key-world-energy-statistics-2021 (accessed on 16 May 2022).
[2] Ministry of Businesss, Innovation & Employment Dairy Manufacturing Fact Sheet, 2016.
[3] T.G. Walmsley, M.J. Atkins, M.R.W. Walmsley, J.R. Neale, Appropriate Placement of Vapour Recompression in Ultra-low Energy Industrial Milk Evaporation
Systems Using Pinch Analysis, 2016, https://doi.org/10.1016/j.energy.2016.04.026.
[4] S. Boldyryev, A.A. Shamraev, E.O. Shamraeva, The retrofit of the calcium chloride production by Pinch approach and process modifications, Appl. Therm. Eng.
189 (2021) 116775, https://doi.org/10.1016/j.applthermaleng.2021.116775.
[5] W.H. Park, Y. Gwak, I.J. Choi, J.-W. Kim, I. Lee, C. Park, J.-K. Kim, Process integration and design of amino acid purification processes for energy savings: an
industrial case study, Kor. J. Chem. Eng. (2024), https://doi.org/10.1007/s11814-024-00172-8.
[6] A. Tsochatzidi, A.L. Arvanitidis, M.C. Georgiadis, Model based optimization of energy consumption in milk evaporators, Processes 12 (2024) 209, https://
doi.org/10.3390/pr12010209.
[7] J.J. Klemeš, P.S. Varbanov, T.G. Walmsley, X. Jia, New directions in the implementation of Pinch methodology (PM), Renew. Sustain. Energy Rev. 98 (2018)
439–468, https://doi.org/10.1016/j.rser.2018.09.030.
[8] MBIE Energy in New Zealand 2020, 2020, p. 76.
[9] F. Bühler, B. Zühlsdorf, T.-V. Nguyen, B. Elmegaard, A comparative assessment of electrification strategies for industrial sites: case of milk powder production,
Appl. Energy 250 (2019) 1383–1401, https://doi.org/10.1016/j.apenergy.2019.05.071.
[10] J. Atuonwu, S. Tassou, Decarbonisation of food manufacturing by the electrification of heat: a review of developments, technology options and future directions,
Trends Food Sci. Technol. 107 (2021) 168–182, https://doi.org/10.1016/j.tifs.2020.10.011.
[11] B.J. Lincoln, L. Kong, A.M. Pineda, T.G. Walmsley, Process integration and electrification for efficient milk evaporation systems, SSRN (2021). https://
papers.ssrn.com/abstract=3995404.
[12] J.-K. Kim, H. Son, S. Yun, Heat integration of power-to-heat technologies: case studies on heat recovery systems subject to electrified heating, J. Clean. Prod. 331
(2022) 130002, https://doi.org/10.1016/j.jclepro.2021.130002.
[13] C. Cui, N.V.D. Long, J. Sun, M. Lee, Electrical-driven self-heat recuperative pressure-swing azeotropic distillation to minimize process cost and CO2 emission:
process electrification and simultaneous optimization, Energy 195 (2020) 116998, https://doi.org/10.1016/j.energy.2020.116998.
[14] J.-C. Bonhivers, A. Moussavi, A. Alva-Argaez, P.R. Stuart, Linking Pinch analysis and bridge analysis to save energy by heat-exchanger network retrofit, Appl.
Therm. Eng. 106 (2016) 443–472, https://doi.org/10.1016/j.applthermaleng.2016.05.174.
[15] M.R.W. Walmsley, N.S. Lal, T.G. Walmsley, M.J. Atkins, A Modified Energy Transfer Diagram for Heat Exchanger Network Retrofit Bridge Analysis, vol. 61,
2017, pp. 907–912, https://doi.org/10.3303/CET1761149.
[16] N.S. Lal, T.G. Walmsley, M.R.W. Walmsley, M.J. Atkins, J.R. Neale, A novel heat exchanger network bridge retrofit method using the modified energy transfer
diagram, Energy 155 (2018) 190–204, https://doi.org/10.1016/j.energy.2018.05.019.
[17] T.G. Walmsley, N.S. Lal, P.S. Varbanov, J.J. Klemeš, Relating Bridge Analysis for Heat Exchanger Network Retrofit Identification to Retrofit Design, vol. 70,
2018, pp. 295–300, https://doi.org/10.3303/CET1870050.

12
B.J. Lincoln et al. Case Studies in Thermal Engineering 60 (2024) 104601

[18] F. Schlosser, H. Wiebe, T.G. Walmsley, M.J. Atkins, M.R.W. Walmsley, J. Hesselbach, Heat pump bridge analysis using the modified energy transfer diagram,
Energies 14 (2021) 137, https://doi.org/10.3390/en14010137.
[19] F. Schlosser, M. Jesper, J. Vogelsang, T.G. Walmsley, C. Arpagaus, J. Hesselbach, Large-scale heat pumps: applications, performance, economic feasibility and
industrial integration, Renew. Sustain. Energy Rev. 133 (2020) 110219, https://doi.org/10.1016/j.rser.2020.110219.
[20] F. Schlosser, S. Zysk, T.G. Walmsley, L. Kong, B. Zühlsdorf, H. Meschede, Break-even of high-temperature heat pump integration for milk spray drying, Energy
Convers. Manag. 291 (2023) 117304, https://doi.org/10.1016/j.enconman.2023.117304.
[21] Walmsley, T. Open Pinch.
[22] Aspentech Aspen HYSYS | Process Simulation Software | AspenTech Available online: https://www.aspentech.com/en/products/engineering/aspen-hysys
(accessed on 7 June 2022).
[23] Piller Blowers & Compressors Customized MVR Blower: VapoFlex | Piller Blowers & Compressors Available online: https://www.piller.de/products-
applications/mvr-blower/vapoflex/(accessed on 13 May 2022).

13