Temere is a predictive microbiology application for exploring how microbial concentration changes during food storage and processing. You can build a sequence of process stages, predict growth or inactivation and investigate how variation in product properties and processing conditions affects the results.
Working with a scenario
A scenario describes the product and the processes you want to investigate. Select an organism and food matrix, define the starting product properties and add process stages such as storage, transport or heat treatment. Each stage is represented by a step module with its own models and inputs.
The predicted concentration at the end of each included module becomes the starting concentration for the next. This lets you assess the combined effect of the process sequence.
A typical workflow is:
Define the system: select the organism, food matrix and an environmental factor where available.
Set the inputs: enter product properties and processing conditions using single values, distributions or file data.
Build the process sequence: add the required modules and arrange them in the order in which the processes occur.
Preview and simulate: open Scenario Preview to inspect the predicted concentration over time. When one or more parameters are set to something other than a single value, the preview runs a short simulation to show variability. Use Simulations to run Monte Carlo simulations, which repeatedly sample from your input distributions or data to calculate a range of possible outcomes.
Explore the results: review concentration changes and the percentage of simulated products above your chosen level of concern.
Available tools
Simulation tools use the latest successful simulation results. Sampling Plan can also use a custom batch:
Time Optimisation: estimates the maximum shelf life or minimum heat treatment time that keeps the simulated percentage of products above the level of concern within your selected target.
Dose-response: estimates the probability of illness and annual number of cases following risk assessment principles using the final simulated concentration, a selected dose-response model and the serving and consumption inputs.
Sampling Plan: estimates the probability that a simulated or custom batch would be rejected under a selected regulatory or custom sampling plan.
Sensitivity Analysis: identifies variable inputs most strongly associated with the final simulated concentration using correlation analysis.
General tools can be used independently of a simulation:
Distribution Fitting: fits probability distributions to numerical data and provides charts and statistics to help you compare the fits.
Model Fitting: estimates primary or secondary model parameters from experimental data and provides charts and statistics to help you evaluate the fit. Fitted models can be stored or downloaded for use in the simulations workflow.
Using this guide
Use Contents to select a topic. Search guide filters the topics using their titles, keywords and text. Use On this page to move between headings within the selected topic.
Predictive Microbiology & Modelling Background
Predictive microbiology uses mathematical models to describe and predict how microorganisms behave under defined environmental and processing conditions. Instead of relying only on individual experimental observations, these models describe general relationships between microbial concentration, time and factors such as temperature, pH, water activity and processing intensity.
Model Types
Temere uses two main classes of predictive models:
Primary models describe how microbial concentration changes with time, for example during growth, survival or thermal inactivation.
Secondary models describe how parameters of a primary model, such as the maximum growth or inactivation rate, change with environmental conditions such as temperature, pH or water activity.
Microbial concentration is generally displayed in Temere as:
\( N(t) = \log_{10} \left( \frac{CFU}{g} \right) \)
The equations below provide a compact overview of the model families available in Temere. The exact parameters, coefficients and valid ranges depend on the selected organism, food matrix, environmental factor and model context.
Primary Models
Primary models describe the change in microbial concentration over time. The appropriate formulation depends on whether the process involves growth, heat treatment or non-thermal inactivation and survival.
Growth
Baranyi & Roberts growth model
A mechanistic sigmoidal growth model that accounts for the initial physiological state of the microbial population, i.e. the cell history, the transition into exponential growth and the approach to a maximum population density.
A canonical form is:
\[ y(t) = y_{0} + \mu_{max} A(t) - \ln{ \left( 1 + \frac{e^{\mu_{max} A(t)} - 1}{ e^{y_{max} - y_{0}} } \right) } \]
with the adjustment function:
\[ A(t) = t + \frac{1}{\mu_{max}} \ln{ \left( e^{-\mu_{max} t } + e^{-h_{0}} - e^{-\mu_{max} t - h_{0}} \right) } \]
where \( y_{0} \) is the initial population, \(y_{max}\) is the maximum population density, \(\mu_{max}\) is the maximum specific growth rate and \(h_{0}\) describes the initial physiological state.
Buchanan model
A simple piecewise model that separates microbial growth into lag, exponential and stationary phases.
\[
N(t)=
\begin{cases}
N_0, & t \le \lambda \\[4pt]
N_0+\mu_{\mathrm{max}}(t-\lambda), & \lambda < t < t_{\mathrm{max}} \\[4pt]
N_{\mathrm{max}}, & t \ge t_{\mathrm{max}}
\end{cases}
\]
where \(\lambda\) is the lag time and \(t_{max}\) is the time at which the maximum population is reached.
In Temere, the lag phase can be linked to the initial physiological state of the microbial population rather than being specified independently. This is represented by the dimensionless physiological-state parameter \(\alpha_{0}\), with:
\[ 0 < \alpha_{0} \le 1 \]
Values of \(\alpha_{0}\) close to 1 represent cells that are well adapted to the current environment and therefore have little or no lag, whereas smaller values represent a less adapted initial population and result in a longer lag phase.
For the Buchanan model, Temere converts \( \alpha_{0} \) into a lag time according to:
\[
\lambda = - \frac{ \log_{10}{ \left( \alpha_{0} \right) } }{ \mu_{max} }
\]
where μ_max is expressed in \( \log_{10}{ \left( \frac{CFU}{g} \right) } \) per unit time.
Consequently, the estimated lag phase depends jointly on the physiological state of the cells and their predicted growth rate under the selected environmental conditions.
This approach allows changes in environmental conditions that affect \( \mu_{max} \) to also influence the duration of the predicted lag phase.
Modified Gompertz model
An empirical sigmoidal model widely used to describe microbial growth curves. Its parameters can be expressed in biologically interpretable terms such as lag time, maximum growth rate and maximum population density.
\[
N(t) = N_0 + (N_{\mathrm{max}} - N_0)
\exp\left\{
-\exp\left[
\frac{\mu_{\mathrm{max}}e}{N_{\mathrm{max}} - N_0}
(\lambda - t) + 1
\right]
\right\}
\]
Modified Logistic model
An empirical sigmoidal growth model based on the logistic curve. It provides a smooth transition between the initial, exponential and stationary phases.
\[
N(t) = N_0 +
\frac{N_{\mathrm{max}} - N_0}
{1 + \exp\left[
\frac{4\mu_{\mathrm{max}}}{N_{\mathrm{max}} - N_0}(\lambda - t) + 2
\right]}
\]
Thermal inactivation
Baranyi & Roberts thermal inactivation model
Applies the Baranyi framework to thermal inactivation by using a negative maximum rate. It can represent non-linear inactivation behaviour, including an initial shoulder before the main decline phase.
\[
\mu_{max} < 0
\]
Bigelow model
The classical log-linear thermal inactivation model. At a constant temperature, the microbial population decreases by one log unit for every decimal reduction time, \(D_{T}\).
\[
N_{t} = N_{0} - \frac{t}{D_{T}}
\]
where \(D_{T}\) is the time required at temperature \(T\) to obtain a \(1-\log_{10}\) reduction.
D-value relationship
When Temere expresses the inactivation rate \(\mu_{max}\) as a negative slope in \( \log_{10} \frac{CFU}{g} \) per hour and the D-value in minutes, the relationship is:
\[
D = - \frac{60}{\mu_{max}}
\]
with:
\[
\mu_{max} < 0 \left[ \frac{\log_{10}\left( \frac{CFU}{g} \right)}{h} \right], D \left[ \text{min} \right]
\]
Non-thermal inactivation & survival
Baranyi & Roberts survival model
Uses the Baranyi framework to describe population decline rather than growth. The same general structure is retained, but the kinetic rate is negative.
\[
\mu_{max} < 0
\]
The resulting curve can represent an initial shoulder or adaptation period followed by a decline in microbial concentration.
Secondary Models
Secondary models describe the effects of environmental or processing conditions on primary-model parameters, such as growth and inactivation rates or decimal reduction times.
Growth
Ratkowsky square-root model
Relates microbial growth rate to temperature above a theoretical minimum temperature for growth.
\[
\sqrt{\mu_{max}} = b \left( T - T_{min} \right)
\]
where \(b\) is an empirical coefficient and \(T_{min}\) is the theoretical minimum temperature for growth.
Extended square-root model
Extends the Ratkowsky square-root relationship to account for both lower and upper temperature limits.
\[
\sqrt{\mu_{max}} = b \left( T - T_{min} \right) \left[ 1 - \exp{ \left( c \left( T - T_{max} \right) \right) } \right]
\]
where \(T_{max}\) is the upper temperature limit and \(b\) and \(c\) are fitted coefficients.
Arrhenius model
Describes the temperature dependence of a kinetic rate using an activation-energy relationship.
\[
\mu = \mu_{ref} \exp{ \left[ - \frac{E_{\alpha}}{R} \left( \frac{1}{T} - \frac{1}{T_{ref}} \right) \right] }
\]
where \( E_{\alpha} \) is the activation energy, \( R \) is the universal gas constant and \( T \) and \( T_{ref} \) are absolute temperatures.
Polynomial growth model
An empirical response-surface model that can describe the effect of one or more environmental variables and their interactions on a growth parameter.
\[
y = \beta_{0} + \sum_{i}{\beta_{i} x_{i}} + \sum_{i}{\beta_{ii} x_{i}^{2}} + \sum_{ i < j }{\beta_{ij} x_{i} x_{j}}
\]
The active linear, quadratic and interaction terms depend on the model definition and the available environmental inputs.
Cardinal model
The cardinal model describes the effect of environmental conditions on microbial growth using biologically meaningful minimum, optimum and maximum values.
Cardinal models express growth relative to the conditions at which growth becomes impossible or reaches its optimum.
For temperature, a cardinal response can be expressed generally as:
\[
\mu_{max} \left( T \right) = \mu_{opt}, \gamma_{T} \left( T \right)
\]
where \( \mu_{opt} \) is the growth rate under optimum conditions and \( \gamma_{T} \left( T \right) \) is a dimensionless temperature response function defined by the cardinal temperatures \(T_{min}\), \(T_{opt}\) and \(T_{max}\) with:
\[
0 \le \gamma_{T} \left( T \right) \le 1
\]
Within the growth domain, \(\gamma_{T} \left( T \right)\) increases from zero at \(T_{min}\), reaches 1 at \(T_{opt}\) and decreases towards zero at \(T_{max}\).
The same cardinal principle can be applied to other environmental factors, such as pH and water activity, using their corresponding minimum, optimum and maximum values.
When several environmental factors are considered, their effects may be combined through multiplicative \(\gamma\)-terms and a multiplicative interaction term \( \xi \) quantifying the combined effect of said parameters:
\[
\mu_{opt} = \gamma_{T} \left( T \right) \times \gamma_{pH} \left( pH \right) \times \gamma_{a_{w}} \left( a_w \right) \times \xi
\]
The exact cardinal formulation, parameters and environmental factors used in Temere depend on the selected model and available organism-specific coefficients.
Thermal inactivation
z-value model
Describes the temperature dependence of the D-value. The z-value is the temperature increase required to reduce the D-value by a factor of 10.
\[
\log_{10}{D_T} = \log_{10}{D_{ref} - \frac{T - T_{ref}}{z}}
\]
where \( D_{ref} \) is the D-value at the reference temperature \( T_{ref} \).
Polynomial thermal inactivation model
An empirical secondary model that relates an inactivation parameter to temperature or other processing variables.
\[
y = \beta_{0} + \sum_{i}{\beta_{i} x_{i}} + \sum_{i}{\beta_{ii} x_{i}^{2}} + \sum_{ i < j }{\beta_{ij} x_{i} x_{j}}
\]
Non-thermal inactivation & survival
Polynomial survival model
An empirical secondary model that relates a survival or inactivation rate to one or more environmental conditions.
\[
y = \beta_{0} + \sum_{i}{\beta_{i} x_{i}} + \sum_{i}{\beta_{ii} x_{i}^{2}} + \sum_{ i < j }{\beta_{ij} x_{i} x_{j}}
\]
How models are used in Temere
A predictive model should always be interpreted within the conditions for which it was developed. Selecting a mathematical model does not by itself establish that it is appropriate for a particular food, microorganism or process.
Temere therefore links model availability, coefficients, parameter bounds and required inputs to the selected organism, food matrix, environmental factor and process type. Primary and secondary models can then be combined within individual process modules, while the predicted concentration is propagated sequentially through the complete process chain.
Important
Predictive models are simplifications of microbial behaviour. Predictions should be interpreted within the model's applicable domain and alongside microbiological knowledge, experimental evidence and appropriate food-safety requirements.
Core Concepts
A Temere scenario combines a system definition, product properties and a sequence of step modules. Together these describe what you are modelling, the starting conditions and the processes that affect microbial concentration.
System definition
The system definition consists of the selected organism, food matrix and an environmental factor (e.g. nitrites, CO2) where available. These selections determine which models and inputs are available for your scenario.
The selected system should represent the organism and product under consideration. The available input ranges reflect the selected system and models.
Product and process parameters
Product parameters describe the starting product, including its initial microbial concentration and properties such as pH and water activity. Product properties are carried across the simulation and used throughout the process sequence. When these parameters use distributions or static file data, a value is sampled once for each simulated product and retained across the modules. Microbial concentration changes as each module is evaluated.
Process parameters describe the conditions within an individual step module, such as its duration and temperature. This allows different stages to use different processing conditions.
The available input modes depend on the parameter: single value, probability distribution or file data. Single values describe fixed conditions. Distributions and file data allow you to represent variation in the inputs.
Model coefficients
Model coefficients define the behaviour of the selected model. They differ from scenario inputs: temperature describes a processing condition, for example, while a model coefficient describes how the organism responds to that condition.
Some coefficients are fixed for the selected model and system. Others can be edited within the module. Coefficients that must remain in a defined order are grouped in a shared control that preserves that order, such as minimum, optimum and maximum values.
Changing a coefficient changes the model's response. Use values supported by the model source or your experimental data.
Linked process stages
Step modules represent successive stages in the same scenario. Their order determines the sequence in which the models are applied.
The concentration predicted at the end of one included module becomes the starting concentration for the next. In Monte Carlo simulations, each simulated product follows this sequence, retaining the effects of the preceding stages.
Excluding a module removes its effect from the calculation. The next included module receives the concentration from the preceding included stage.
Step modules
Building Scenarios
Build and edit scenarios in the Inputs panel, available in Scenario Preview and Simulations. Define the system first, set the product parameters and add the process stages in the order in which they occur.
Define the system
Under System definition, select:
Organism: select the microorganism to be modelled. Organism names include strain information where applicable.
Food matrix: only broth models are currently included.
Environmental factor: select the inhibiting factor present in the food, nitrites as a preservative or CO2 in the product packaging. Choose None if no inhibiting factor is present.
The environmental factors depend on the selected organism and matrix. The system definition also determines the available product parameters, process types and models.
Changing the system definition resets the model inputs and step modules. If you have edited the scenario, Temere asks you to confirm the change. Save the scenario first if you want to retain its current configuration.
Set the product parameters
Review the product parameter cards below the system definition. These include initial microbial concentration, physiological state, pH, water activity and inputs associated with the selected environmental factor.
For each parameter:
Check the units displayed on the card.
Select the input mode: single value, distribution or file.
Enter the value or provide the distribution settings or file data required by that mode.
The active mode determines which input is used in the calculation. Check every card rather than relying on its default value.
Product parameters are carried across the simulation. For distributions and static file inputs, each simulated product receives a sampled value that is retained throughout the process sequence. Initial concentration describes the start of the sequence; subsequent concentrations are calculated by the step modules.
See Parameter Cards for instructions on distributions, file formats and time-varying profiles.
Add process stages
Select the + button below the module list to open the Add module dialog.
Choose the process Type:
Growth
Thermal inactivation
Non-thermal survival
Disabled process types are unavailable for the selected system.
Choose a stage from Predefined modules or select Custom and enter a module name. Predefined modules provide named stages such as storage, transport and cooking. Select the primary and secondary models for the stage, then select Add.
Repeat this process for each stage in the scenario. Separate modules allow you to describe stages with different durations, temperatures or other processing conditions.
Set the process parameters
Review the parameter cards in each added module. Enter the duration and other required conditions using the input modes available on each card.
Use separate modules for distinct stages with their own conditions. For example, industrial storage, transport and retail storage can each have a different duration and temperature.
For conditions that change during a stage, use a dynamic file profile on parameters that support it. The uploaded timeline determines the module’s duration. Dynamic profiles increase computation time because the model evaluates the changing conditions across the stage.
Review editable model coefficients within the module. These control the model’s response and should reflect the model source or experimental data used for the scenario.
Use a fitted model
Models stored through Model Fitting are available for compatible systems and process types. To use a downloaded fitted model, select Import fitted model in the Add module dialog and choose the model file.
Select the fitted model from the secondary model list when building the module. Its availability depends on the organism, food matrix, environmental factor and process type specified when the model was stored.
Arrange and include modules
Modules are added at the end of the process sequence. If you add a stage that belongs between existing modules, drag it into the required position. Concentration is passed from one included module to the next, so changing the order changes the calculation.
Use the module’s toggle switch to exclude it temporarily while retaining its inputs. Excluded modules do not contribute to the preview or simulation.
Compatibility checks
Review the parameter validity indicator and any warnings before running:
Red indicators identify invalid or missing inputs that must be corrected.
Yellow indicators identify values outside the model’s usual range but within physical limits. Review these values and the model’s suitability before proceeding.
Blue module indicators identify modules that use dynamic profiles, with duration determined by the uploaded timeline.
Use the validity indicator to locate the affected inputs.
Table input validation, warnings and errorsModel compatibility range checkHint of a module's time being bypassed due to a dynamic profile being active
The Inputs panel is shared between Scenario Preview and Simulations. You can switch between the two while building the scenario to see its general behaviour before running the full Monte Carlo simulations.
Once the scenario is ready, open Simulations, set the number of simulated product units and the level of concern and click Run simulations. Use the seed settings to reproduce simulations with the same scenario and inputs. See Settings and Exports for details.
After changing the scenario, run the simulations again before using Simulations tools so that those tools use the updated results.
Save and load scenarios
Use Save current scenario at the top of the Inputs panel to download the scenario as a JSON file. This preserves the system definition, parameter inputs, module order, inclusion settings and model coefficients.
Use Load scenario from file to restore a saved scenario. Loading replaces the current scenario, so save any configuration you want to retain first.
A scenario file stores the setup. Run the simulations after loading it to generate results for Simulations tools.
Save and load scenario buttons
Parameter Cards
Parameter cards provide a consistent and flexible way to define model inputs across Temere. Each card displays the parameter name, units and active input mode. Depending on the parameter (initial concentration, physiological state, pH, water activity, environmental factor, temperature, time, etc), you can enter a single value, define a probability distribution or supply data through a table or file.
Using a parameter card
Click the card heading to expand or collapse its inputs. Use the three mode buttons to choose:
Single value mode: use one fixed value.
Distribution mode: sample values from a probability distribution.
File mode: use static data or a dynamic profile.
The active mode determines which input is used in calculations. Check the displayed units before entering values.
The footer contains Quick tips for the parameter and controls to reset the active mode or expand the card into a larger view. Resetting single value or distribution inputs restores their defaults. Resetting file inputs clears the data and fitted distributions.
Single value input
Enter a value directly in the number field or adjust it using the slider. The same value is used for every simulated product.
The slider provides a convenient way to adjust the input within its displayed range. Model-valid ranges are shown separately in the footer so you can check whether the value is appropriate for all modules that use it.
Distribution input
Select a distribution and enter its parameters. The available distributions depend on the parameter being modelled. The chart displays the probability density and updates as you change the distribution settings.
Set the lower and upper truncation limits to restrict sampling to a defined range. Truncation is useful for excluding extreme tail values that are physically implausible or fall outside the conditions supported by the selected models. Only values within these limits are sampled.
For product parameters, a value is sampled once for each simulated product and carried across the process sequence. Process parameters are sampled for the module in which they are defined.
Discrete distributions
Use a discrete distribution when an input has a defined set of possible values. Enter each value and its probability in the table or upload a two-column value–probability file.
Probabilities are treated as relative weights. If they do not total 100%, Temere normalises them for sampling. Weights must be non-negative and at least one must be greater than zero.
Poisson initial concentration
The Poisson option is available for initial microbial concentration. Enter the concentration in CFU/g and product mass in g. Temere uses these inputs to sample microbial counts and calculate the corresponding concentration.
Check the units carefully because this option uses CFU/g rather than the logarithmic concentration used elsewhere.
File input
Enter values directly into the table, paste them from a spreadsheet or upload a CSV, XLSX or XLS file. Uploaded data appear in the table for review and editing. Files can include a header row.
If you edit the data in the table, use its download controls to save the revised data as CSV or XLSX.
Static data
Static data contain one column of numeric parameter values. During Monte Carlo simulations, values are sampled from this column with replacement, meaning the same recorded value can be selected more than once.
The histogram shows the distribution of the entered data. Static data represent variation between simulated products rather than changes over time within a stage.
Fit a distribution to static data
With at least five numeric values, enable Fit distribution to fit candidate distributions to the data.
Use Fit criterion to choose how successful fits are ranked. Select a result from Fitted distribution to view its curve and fitted parameters. You can choose any successful fit, regardless of its ranking.
When fitting is enabled and a fitted distribution is selected, simulations sample from that distribution. With fitting disabled, simulations sample directly from the table values.
See Distribution Fitting for more detail on fit criteria and interpretation.
Dynamic profiles
For parameters that support time-varying inputs, select Dynamic and provide two columns: time and parameter value.
The profile must:
Contain at least two complete numeric rows.
Start at time zero.
Use non-negative time values.
Contain at least two distinct time points.
The chart shows how the parameter changes over time. Temere calculates intermediate values between the supplied points and evaluates the changing conditions across the stage.
The uploaded timeline determines the module's duration, replacing the duration entered on its time card. Dynamic profiles increase computation time because the model evaluates conditions throughout the profile.
Model-valid ranges
Product parameters are shared across modules, but the models in those modules can have different valid ranges.
Expand Quick tips in the card footer to view Model-valid ranges. The coloured bars and labels show the range for each model and identify its module. The Common zone is the overlap between these ranges: values within it fall inside the stated range of every model shown.
Model range warning in a parameter card
For example, if one model supports pH 3.90-7.40 and another supports pH 4.00-7.10, the common zone is pH 4.00-7.10. A value of 3.90 is within the first model’s range but outside the second, so the card highlights it for review.
For distribution inputs, review the truncation limits against the common zone. For file inputs, review the supplied values. If the model ranges do not overlap, the card reports No common valid zone across active steps.
Click the card's validity indicator to open the relevant footer information. Model-valid ranges describe the conditions supported by the models; physical limits define which input values the application accepts.
Values outside the common zone trigger a warning but do not prevent simulations from running, provided they remain within physical limits and meet the other input requirements. Predictions outside a model’s stated range represent extrapolation and should be interpreted accordingly.
Warnings and invalid inputs
Yellow highlighting identifies inputs outside the model-valid range that require review. These warnings allow calculations to continue.
Invalid inputs, including values outside physical limits, incomplete data or missing required file inputs, must be corrected before simulations can run. Review the highlighted fields or table cells and the accompanying messages.
A blue module indicator identifies a dynamic profile whose timeline determines the module's duration.
Table input validation, warnings and errorsModel compatibility range checkHint of a module's time being bypassed due to a dynamic profile being active
Step Modules
Step modules represent individual stages in the process sequence. Each module contains its process parameters, selected models and editable model coefficients. Its toggle switch controls whether it contributes to the scenario calculations.
Module colours
Modules are colour coded by process type:
Growth
Thermal inactivation
Non-thermal survival
Opening and editing a module
The module heading displays its name and the selected primary and secondary models. Click the heading to expand or collapse the module and access its inputs.
See Parameter Cards for instructions on entering process conditions and Core Concepts for the distinction between parameters and model coefficients.
Collapsing a module changes only its display. It remains included in calculations while its toggle switch is on.
Including or excluding a module
Use the module’s toggle switch to include or exclude it from Scenario Preview and Simulations.
An excluded module retains its inputs but does not affect the calculated concentration. Switch it on again to include it in the scenario, then click Run simulations to generate results that include its effect.
Changing module order
Modules are added at the end of the process sequence. To move a module, drag the handle at the top of its card to the required position.
Calculations follow the order of the included modules read from top to bottom, with each stage receiving the concentration predicted by the preceding stage.
Module actions
Click the three-dot Module actions button to access:
Collapse inputs: collapses the parameter and coefficient controls without changing their values, keeping the interface compact when working with several modules.
Delete module: removes the module from the scenario.
After deleting a module, click Undo in the deletion notification to restore it. Hovering over the notification pauses its timer.
Preview
Preview shows the predicted change in microbial concentration over time across the included step modules. The shared Inputs panel lets you adjust the scenario and see its general behaviour and possible outcomes before running the full Monte Carlo simulations.
When all inputs are set to single value, the scenario preview is deterministic: the models calculate one concentration curve using those fixed inputs. When distributions or static file data are used, i.e. stochastic method, a short simulation samples the inputs to calculate a median curve and a scenario range based on their 5th and 95th percentiles.
User interface of the Preview tab.
Growth range plot
The Growth range plot displays microbial concentration in log CFU/g against cumulative process time. Vertical boundaries separate the included modules so you can see how each stage contributes to the predicted change.
The curve follows the process sequence, carrying the predicted concentration from the end of each module into the next.
With variable inputs, the solid curve represents the median scenario and the shaded band shows the range calculated from the 5th and 95th percentile inputs. Use Simulations for the full distribution of outcomes and the percentage of products above the level of concern.
Navigating the chart
Use the navigator below the main chart to focus on part of the process sequence. Drag its handles to adjust the visible time frame.
To zoom, hold Shift while scrolling on a PC or pinch on a mobile device. Drag the main chart to pan. Double-click the chart to reset the view to the full process sequence.
Chart details
The Chart details footer summarises each included module, including its duration, starting and final concentration, concentration change and predicted rate.
Use these values alongside the curve to compare the effects of individual stages.
Preview status
The status bar reports progress, the final predicted concentration and total process time. It also displays warnings when an input cannot be used as configured.
Review any warning before interpreting the curve. If a variable input cannot be sampled, the preview uses its current single or default value and reports the substitution.
Status bar of the growth range chart in the Preview tab.
Simulations
The Simulations tab runs stochastic Monte Carlo simulations for the scenario defined in the shared Inputs panel. Each iteration represents one simulated product passing through the included process stages.
Inputs defined by distributions or static data are sampled to represent variability. Single value inputs remain fixed and dynamic profiles describe conditions that change over time. The results show the resulting distribution of microbial concentrations after each stage.
User interface of the Simulations tab.
Simulation setup
Set Simulated product units and Level of concern (LoC) in the Simulation setup card before running.
Simulated product units
This sets the number of Monte Carlo iterations. Enter a whole number between 2 and 1,000,000.
Increasing the number of simulated products generally produces more stable estimates of summary statistics and the percentage above the level of concern, but increases computation time.
Note
Increasing the number of simulated products does not improve the suitability of the selected models or inputs.
Scenarios with dynamic profiles take longer because the models evaluate changing conditions throughout each stage.
Level of concern
Enter the concentration threshold in log CFU/g. This value is used to calculate the number and percentage of simulated products above the threshold and to highlight results in the charts.
The level of concern is a comparison threshold; it does not change the predicted microbial concentrations.
Running simulations
Click Run simulations to start. Temere checks the inputs, samples the variable parameters and evaluates the included modules in their process order.
Product parameters are sampled once for each simulated product and carried through the sequence. Each module uses its own process inputs and passes its predicted concentration to the next included module.
The progress display in the Simulation setup footer identifies the active stage and reports completion or failure.
If a run stops, review the reported message and correct the affected input before running again. For example, sampling can fail when truncation limits retain an extremely small part of a distribution. Adjust the distribution settings or limits to represent the intended input range.
Reproducing simulations
Monte Carlo results can differ between runs because the inputs are sampled randomly. Use the seed settings, accessed at the bottom of the sidebar, to reproduce a simulation with the same scenario, inputs and number of simulated products.
See Settings and Exports for details.
Results and simulation tools
After a successful run, the results appear in Simulation summary, Concentration summary and Simulation results per module. See Reading Results for explanations of the statistics, charts and AI-assisted interpretation.
The latest successful simulation provides the results used by Simulations tools:
Time Optimisation
Dose-response
Sampling Plan
Sensitivity Analysis
Editing the scenario does not recalculate its results. Click Run simulations again to generate results for the revised scenario. Each successful run replaces the previous simulation results. Run the relevant Simulations tools again to update their analyses using the new results.
Reading Results
The results in Simulations describe the initial microbial concentration and the concentration after each included module. Summary statistics and charts show how concentration changes through the process sequence and how outcomes vary between simulated products.
Concentrations are displayed in log CFU/g. Results describe the simulated scenario under its selected models and inputs.
Simulation summary
The Simulation summary table provides two views:
Concentration: summarises the initial concentration and the concentration after each included module.
Rate: summarises the predicted growth or inactivation rates within each module, expressed in log(CFU/g)/h. For dynamic profiles, each simulated product’s rate is averaged over the stage with weighting by interval duration.
The concentration view reports:
Min and Max: the lowest and highest simulated concentrations.
Q1, Median and Q3: the 25th, 50th and 75th percentiles.
Mean: the average simulated concentration on the logarithmic scale.
SD: the standard deviation.
IQR: the difference between Q3 and Q1, describing the spread of the central 50% of results.
> LoC: the number and percentage of simulated products above the level of concern.
Use Show indicators to display or hide arrows showing increases or decreases relative to the preceding stage. These compare the displayed statistics rather than individual simulated products.
Level of concern
The level of concern is the concentration threshold selected for the simulation. The > LoC results count simulated products whose concentration is strictly above that threshold.
The percentage above the level of concern describes concentration exceedance.
Concentration summary
The Concentration summary compares the initial concentration with the results after each included module. Choose between Boxplot and Ridgeline.
In Boxplot view, the box extends from Q1 to Q3 and the line inside marks the median. Whiskers extend to the most extreme results within 1.5 times the IQR below Q1 or above Q3. Results beyond the whiskers are shown as dots representing outliers.
In Ridgeline view, stacked density curves show the shape and spread of each concentration distribution. This view helps compare shifts in concentration, asymmetry and multiple peaks across stages.
Simulation results per module
Each result has a histogram and a gauge. The histogram shows the concentration distribution, while the gauge reports the percentage of simulated products above the level of concern.
Choose the histogram Scale:
Frequency: shows the number of simulated products in each concentration interval.
Density: shows probability density, allowing distribution shapes to be compared independently of the number of simulated products.
Use Show LoC to display or hide the threshold marker and associated bar colouring. Bars above the threshold are highlighted in red. A bar containing the threshold is highlighted in yellow because its interval spans concentrations on both sides.
The gauge is calculated from individual simulated concentrations.
Comparing and navigating charts
Enable Sync charts to use common axes and synchronise zooming across the module histograms. This makes changes in concentration and spread easier to compare.
With synchronisation off, each histogram can be explored independently.
Hold Shift while scrolling to zoom and drag left or right to pan. Double-click to reset the view.
AI-assisted interpretation
AI interpretation button in the Simulation Summary card
After a successful simulation, click Interpret results in the Simulation summary card to request an AI-assisted explanation.
The interpretation discusses the overall results, changes between stages, exceedance of the level of concern and variability. It uses summary statistics together with the scenario inputs and selected models.
The request sends summaries and information about the models and input settings. It does not send individual Monte Carlo results, uploaded files or chart images. Review the explanation alongside the tables and charts.
Note
AI-assisted interpretation supports your assessment of the results; it does not validate the models or establish regulatory compliance.
Dose-response
Dose-response estimates the probability of illness per contaminated serving and the expected annual number of cases. It uses the final microbial concentration from the latest successful stochastic simulation together with the selected dose-response model and consumption inputs.
Run the scenario in Simulations before opening this tool. The organism and concentration results are inherited from that simulation. The analysis uses the concentration after the last included module.
Select a model
Choose the following inputs:
Dose-response model: the model used to calculate illness probability from the ingested dose.
Target population: the population represented by the available model parameters.
Coefficient set: the published model coefficients to use for the selected population.
The available choices depend on the organism and selected model. Coefficients are either fixed values or distributions, depending on the selected source.
Serving and consumption inputs
Set the three parameter cards:
Serving size: the amount consumed on one eating occasion, in grams.
Prevalence: the percentage of products that are contaminated.
Annual eating occasions: the total number of eating occasions per year for the population or consumption setting being assessed.
Each card supports single values, distributions and static file data. See Parameter Cards for instructions on these input modes.
Serving size converts the simulated concentration into an ingested dose:
\[
D = C \times S
\]
where \(D\) is the ingested dose in CFU, \(C\) is the microbial concentration expressed in CFU/g and \(S\) is the serving size expressed in g.
Temere converts the simulation’s logarithmic concentrations to CFU/g before calculating the dose.
Prevalence and annual eating occasions are used to estimate annual cases:
\[
E = P_{\mathrm{ill}} \times \frac{p}{100} \times A
\]
where \(E\) is the expected annual number of cases, \(P_{\mathrm{ill}}\) is the probability of illness per contaminated serving, \(p\) is prevalence expressed as a percentage and \(A\) is the annual number of eating occasions.
The probability of illness describes a contaminated serving. Prevalence accounts for the proportion of eating occasions involving contaminated products when calculating annual cases.
Model preview
The Model preview chart shows the relationship between ingested dose and probability of illness for the selected model.
Fixed coefficients produce a single curve. When coefficients use distributions, the preview shows a median curve and a shaded band spanning the 5th to 95th percentiles of the calculated probabilities.
This chart describes the selected dose-response model. The result charts incorporate the simulated concentrations and the serving and consumption inputs.
Run the analysis
Click Run dose-response after selecting the model and completing the input cards.
The analysis uses the number of simulated products from the latest successful simulation. It samples the dose-response inputs and any distributed coefficients, calculates illness probabilities and estimates annual cases.
After changing these inputs, click Run dose-response again to update the results. If you run a new scenario in Simulations, rerun the dose-response analysis using the new concentration results.
Read the results
The Dose-response results card contains two histograms:
Pill: the distribution of calculated probabilities of illness.
Cases: the distribution of expected annual cases.
Use Grid to display both charts together or Single to enlarge one chart. In Single view, use Chart to choose Pill or Cases.
For Pill scale, choose Linear or Log10. The logarithmic view helps distinguish small probabilities. Zero values are excluded from this view.
The summary table reports the minimum, first quartile, median, mean, third quartile and maximum for probability of illness and expected cases.
Expected cases are modelled averages, so they can include fractional values. They are estimates based on the scenario and consumption assumptions rather than observed case counts.
Sampling Plan
Sampling Plan estimates the probability that a batch would be rejected under a selected regulatory or custom microbiological sampling plan. It repeatedly draws samples from the selected batch and applies the plan’s acceptance criteria.
Use concentrations from the latest successful simulation or define a custom contaminated batch directly in the tool.
Choose the concentration source
Choose one of the following concentration sources:
Simulation results: use contaminated-unit concentrations from a module in the latest successful simulation. The simulation supplies the organism automatically.
Custom batch: define contaminated-unit concentrations in log CFU/g using a single value, probability distribution or uploaded data. Set Contaminated units to the number of values generated or sampled.
When no successful simulation is available, Custom batch is selected automatically. The generated custom batch is used for both the lot preview and sampling calculation.
Select the population
For simulation results, use Module to sample from to select the process stage at which samples are taken. Sampling uses the concentrations predicted at the end of that module.
Set Prevalence to the percentage of contaminated products, between 0.1% and 100%. The selected concentrations represent contaminated products. Temere adds products with zero CFU/g to form a batch with the specified prevalence.
Set Repetitions to the number of times the sampling plan is applied. Enter a whole number between 1 and 10,000. More repetitions generally produce a more stable estimate of rejection probability but increase computation time.
Choose a sampling plan
Under Sampling plan, choose:
Regulation: select an available plan based on Commission Regulation (EC) No 2073/2005 from Regulation sampling plan. Plans are grouped by food category and supply the plan class and acceptance parameters. These parameters are read-only. For a custom batch, select the organism first.
Custom: choose a two-class or three-class plan and enter its acceptance parameters.
The plan parameters are:
\(n\): the number of product units drawn in each repetition.
\(c\): the maximum permitted number of units exceeding \(m\) in a two-class plan or falling between \(m\) and \(M\) in a three-class plan.
\(m\): the concentration limit separating satisfactory results from results that count towards \(c\).
\(M\): the upper concentration limit in a three-class plan.
Enter \(m\) and \(M\) in CFU/g. Temere converts the selected concentrations before applying the plan.
For custom plans, \(n\) must be between 1 and 100 and \(c\) must be a whole number from 0 to \(n-1\). In three-class plans, \(M\) must be greater than \(m\).
Two-class plans
A two-class plan divides sampled units into satisfactory and unsatisfactory results using \(m\).
The batch is rejected when the number of unsatisfactory units exceeds \(c\). Custom plans count concentrations strictly above \(m\). Regulatory plans use the comparison specified by the selected criterion.
Three-class plans
A three-class plan distinguishes:
Satisfactory: concentration at or below \(m\).
Marginal: concentration above \(m\) and at or below \(M\).
Unacceptable: concentration above \(M\).
The batch is rejected if more than \(c\) sampled units are marginal or if any sampled unit exceeds \(M\).
Lot preview
The Lot preview displays the concentration distribution for contaminated products and identifies the contaminated and uncontaminated portions of the selected batch.
Use it to review the selected module or generated custom batch and prevalence. The plan’s concentration limits are shown alongside the distribution. A custom batch preview appears after sampling begins.
Run the sampling plan
Click Sample to run the selected plan.
Each repetition draws \(n\) products from the selected batch without replacement, so a product cannot be selected twice within the same sample. The complete batch is available again for the next repetition.
The progress display reports population preparation, sampling and result preparation. The sample size cannot exceed the available batch size.
Read the results
The Sampling results card shows:
Rejection probability: the proportion of repetitions in which the plan rejected the batch.
Accepted and rejected counts: the number of repetitions producing each decision.
95% confidence interval: the binomial confidence interval for the estimated rejection probability.
The gauge displays the accepted and rejected proportions.
The bar chart shows the frequency of sampled counts that contribute to \(c\). For two-class plans, these are unsatisfactory units. For three-class plans, these are marginal units; concentrations above \(M\) trigger rejection separately.
The results describe how the selected plan performs for the selected batch. An accepted sample does not establish that every product in the batch meets the concentration limit.
Results produced from simulation concentrations can be included in the simulation report. Custom batch results are not added to that report.
Update the analysis
After changing the concentration source, selected module, custom batch, prevalence or plan settings, click Sample again.
If you run a new scenario in Simulations, rerun Sampling Plan to assess the updated concentration results.
Sensitivity Analysis
Sensitivity Analysis identifies which variable inputs are most strongly associated with the final microbial concentration in the latest successful stochastic simulation.
The analysis compares sampled input values with the concentration after the last included module. It helps identify which sources of variability are associated with higher or lower final concentrations.
Run the scenario in Simulations before opening this tool.
Choose a correlation method
Under Correlation method, select:
Spearman rank correlation: measures whether increasing input values are consistently associated with increasing or decreasing final concentrations. It accommodates monotonic relationships that are not linear and is the default method.
Pearson correlation: measures the strength and direction of a linear relationship between an input and the final concentration.
Run the analysis
Click Run analysis to calculate the correlations.
The analysis includes product and process parameters defined by distributions, static file values or fitted distributions. Single value inputs and dynamic profiles are excluded.
Inputs with no variation in their sampled values are excluded from the analysis.
Read the tornado plot
The Tornado plot ranks inputs by the absolute size of their correlation with final concentration. Longer bars indicate stronger associations.
The direction of each bar shows the relationship:
Positive correlation: higher input values are associated with higher final concentrations.
Negative correlation: higher input values are associated with lower final concentrations.
Correlation close to zero: little association is detected by the selected method.
Overall and grouped views
Under Tornado view, choose:
Overall: ranks all analysed inputs together from strongest to weakest absolute correlation.
Grouped: separates product parameters and module inputs. Product parameters appear first, followed by modules in process order. Inputs are ranked within each group.
Use Overall to compare inputs across the scenario and Grouped to review the inputs associated with each stage.
Results table
The table below the plot identifies each input, its distribution or file input type and its correlation value.
Read the sign together with the magnitude. For example, a correlation of \(-0.8\) represents a stronger association than \(0.3\), with the relationship acting in the opposite direction.
Update the analysis
After changing the correlation method, click Run analysis again.
After a new successful simulation, rerun Sensitivity Analysis to calculate correlations for the updated scenario.
Time Optimisation
Time Optimisation estimates a maximum shelf life or minimum heat treatment time that keeps the simulated percentage of products above the level of concern within a selected target.
The tool uses the scenario, inputs and level of concern from the latest successful stochastic simulation. Run the scenario in Simulations before opening it.
Set the target
Enter Target products > LoC as the maximum acceptable percentage of simulated products above the saved level of concern.
For example, a target of 5% allows no more than 5% of simulated products to exceed that concentration. The recommendation can produce a lower percentage; the target is an upper limit.
To use a different level of concern, change it in Simulations and run the scenario again.
Choose the objective
Under Optimisation objective, choose:
Maximum shelf life: finds the longest duration that meets the target by adjusting selected growth or non-thermal survival stages. Thermal treatment times remain fixed and unaffected.
Minimum heat treatment: finds the shortest duration that meets the target by adjusting one thermal inactivation stage. Other stage durations remain fixed.
The tool changes time inputs. Temperatures, product properties, models and other scenario inputs retain their simulation settings.
Choose the time allocation
Under Time allocation, choose:
Multiple: adjust several eligible module times in the same proportion.
Single: adjust one eligible module time.
For example, reducing times by 20% in Multiple mode reduces every adjustable stage by that proportion. Fixed stages retain their durations.
Minimum heat treatment uses Single mode.
Choose the distribution adjustment
When time inputs use distributions or static data, Distribution adjustment determines how those inputs change:
Shift only: adjusts the central time while retaining the absolute spread where the input supports this.
Entire shape: scales the complete time input, changing both its central value and spread proportionally.
For a normal time distribution, Shift only changes the mean while retaining the standard deviation. Entire shape changes both.
Some fitted distributions require Entire shape. The module displays this requirement when Shift only is unavailable.
For single value inputs, both options scale the entered time.
Select the process window
The window defines the part of the process covered by the optimisation. For Maximum shelf life, the combined duration of the included modules within this window represents the product’s shelf life. Select the stages that make up the shelf-life period you want to assess. For example, final cooking immediately before consumption should sit outside this window because its duration is a preparation step and does not form part of the storage period. Its effect can still be included by selecting it under Measure exceedance after.
For Minimum heat treatment, select a window containing the thermal stage to be adjusted.
Move the two slider handles to select the first and last modules. Every module between them is included. The window is labelled Shelf-life window or Process window, depending on the objective.
For each module within the window, choose:
Fixed: retain the existing time input for stages whose duration is constrained by operating conditions. For example, transport over an established route can have a duration that cannot readily be shortened. Its time remains part of the selected period while adjustable stages, such as storage, are shortened or extended.
Adjustable: allow the tool to change its time input.
Fixed modules remain part of the calculation. Modules using dynamic profiles have fixed durations because their timing comes from the uploaded profile.
Choose where the target is assessed
Use Measure exceedance after to select the stage at which concentrations are compared with the level of concern.
This defaults to the last module in the selected window. You can select a later stage to include the effects of subsequent processing or storage.
Later stages contribute to the calculated concentration but their times remain fixed outside the selected window.
Run the optimisation
Click Optimise shelf life or Optimise treatment, depending on the selected objective.
The tool tests different durations and checks the recommendation using the full number of simulated products from the original simulation. The progress display reports the analysis stages.
The search stays within the permitted bounds of the adjustable time inputs. If the target cannot be met within those bounds, the result reports this. If the target remains satisfied at the time limit, the result identifies that limit.
Read the recommendation
The result compares the current time with the recommended time and reports the resulting number and percentage of products above the level of concern.
For maximum shelf life, the displayed duration covers the selected window, including fixed stages. For minimum heat treatment, it describes the adjustable thermal stage.
The module table lists:
Current time
Recommended time
Input adjustment
For variable time inputs, the displayed duration represents the input’s central value. Review the listed input adjustments to see exactly which values must change.
Boundary chart
The boundary chart plots tested durations against the percentage of products above the level of concern. It highlights the recommendation and current scenario, with reference lines for the target and recommended duration.
Use it to see how the tested durations approach the selected target. The connecting curve shows the trend between tested durations.
Apply the recommendation
The recommendation does not automatically change the scenario. Update the time inputs in the shared Inputs panel using the module table, then click Run simulations.
Apply all listed changes. For example, an Entire shape adjustment to a normal distribution requires changing both the mean and standard deviation.
Use the same seed settings to reproduce the result. With a new random seed, the overall behaviour should remain similar, although the exact percentage above the level of concern can differ because of random sampling.
Distribution Fitting
Distribution Fitting fits probability distributions to numerical data. It provides fitted parameters, goodness-of-fit statistics and diagnostic charts to help you select a distribution that describes the observed variability.
This tool can be used independently of a simulation.
Prepare the data
Enter values directly into the Data input table, paste them from a spreadsheet or upload a CSV, XLSX or XLS file.
For files containing several columns, use Data column to select the variable to fit. Files can include a header row. The selected values appear in the table for review and editing.
At least five numeric values are required and the data must contain variation. Empty cells and non-numeric entries are omitted from fitting. The status message reports how many numeric values will be used and identifies omitted entries.
Use the table controls to add, insert or remove rows and undo or redo edits. If you edit the data, use the download controls to save the revised table as CSV or XLSX.
Changing Data column replaces the table values. If you have edited them, Temere asks you to confirm before discarding those edits.
Fit the distributions
Click Fit distributions to fit the following candidate distributions:
Normal
Logistic
Log-normal
Gamma
Weibull
Exponential
Uniform
Triangular
Beta-PERT
Successful fits appear under Fitted distribution. Select a distribution to view its parameters, statistics and diagnostic charts.
If you change the data, click Fit distributions again to update the results.
Compare and rank fits
Use Fit criterion to rank the successful fits. Fitted distribution lists them from best to worst according to the selected criterion.
The available criteria are:
Akaike information criterion (AIC): compares fit while accounting for the number of estimated parameters. Lower values rank better.
Bayesian information criterion (BIC): also accounts for model complexity, with a stronger penalty for additional parameters as sample size increases. Lower values rank better.
Log-likelihood: measures how well the fitted distribution describes the observed data. Higher values rank better.
Kolmogorov–Smirnov statistic (KS): measures the largest difference between the empirical and fitted cumulative distributions. Lower values rank better.
Cramér–von Mises statistic (CvM): measures differences across the empirical and fitted cumulative distributions. Lower values rank better.
Changing the criterion reorders the existing fits without refitting the data. You can select any successful fit, regardless of its ranking.
Shifted distributions
Some distributions require positive values. When the data include zero or negative values, Temere fits a shifted distribution by applying an offset before fitting.
Shifted fits are labelled and the parameter table reports the shift offset. The charts show the fit on the original data scale.
Diagnostic charts
Use Grid to display all four charts or Single to enlarge one. In Single view, use Chart to select:
Histogram: displays the observed data as a density histogram with fitted probability density curves.
Q-Q: compares observed and fitted quantiles. Alignment with the reference line indicates agreement; departures near the ends reveal differences in the tails.
CDF: compares the empirical and fitted cumulative distribution functions.
P-P: compares empirical and fitted cumulative probabilities. Alignment with the reference line indicates agreement.
Parameters and fit statistics
The results tables display the selected distribution’s parameter estimates, standard errors where available and goodness-of-fit statistics.
Use the parameter estimates to describe the fitted distribution. Standard errors describe uncertainty in those estimates.
Compare fit criteria using the same dataset. A high ranking identifies a comparatively good fit among the successful candidates; the charts help assess how well it describes the observed data.
Use a fit in a scenario
To fit a distribution directly to a scenario input, use File mode on its parameter card and enable Fit distribution. This lets you select the fitted distribution for sampling during simulations.
See Parameter Cards for details.
Model Fitting
Model Fitting estimates model parameters from experimental data. It supports primary and secondary models for growth, thermal inactivation and non-thermal survival.
The tool provides fitted curves, diagnostic charts and statistics to help you evaluate the fit. Fitted secondary models can be stored or downloaded for use in the simulations workflow.
Select a model
Choose the Model type, Process and Model. Click View model to see the equation, observed variables and parameters to estimate. The table columns are generated from the selected model.
For primary models, enter measured concentrations and their corresponding times. For secondary models, enter the measured variable, such as growth or inactivation rate and the corresponding environmental conditions. Check the variables and units shown before entering data.
Prepare and import data
Enter data directly into the table, paste it from a spreadsheet or click Upload data to load a CSV, XLSX or XLS file. In the Match uploaded columns dialog, match your file's columns to those shown for the selected model. Set First row contains column names as appropriate then click Import data.
Each row should contain a measured observation and the corresponding experimental conditions. The column names are generated from the selected model. Enter your measurements under the appropriate names and check the units shown.
For multivariate secondary models, leave an environmental variable's column empty if no measurements are available for it. Temere fits the parameters supported by the supplied data. At least one environmental variable must contain varying numeric values.
Rows with missing or non-numeric measurements in the columns used for fitting are excluded. Click Load example to load example data for the selected model.
Parameters estimated from the data
Temere automatically determines which parameters can be estimated from the available observations and variation in the inputs. For secondary models, varying environmental conditions provide information about their effects. A column containing a constant value does not provide the same information.
For example, if a Gamma Cardinal dataset contains varying temperature and pH values but no water activity observations, Temere estimates the temperature and pH cardinal parameters while omitting the water activity effect.
For polynomial models, linear and quadratic terms are fitted only when their corresponding input values vary across the observations. An interaction term requires every input involved to vary. Terms that do not meet these conditions are omitted from the fit.
Only parameters estimated from your data are shown in the parameter estimates table.
Starting values
Temere automatically selects initial parameter values using the selected model and the range of your data. It tries several starting points during fitting. You do not need to provide initial values.
Run the fit
Click Fit model. The progress display shows validation, fitting and result preparation.
A fit can fail when there are too few complete observations, insufficient variation in the data or values outside the permitted limits. Noisy data or a model that does not adequately describe the observations can also prevent fitting from finding a solution.
Review the displayed message, check the data and selected model then click Fit model again.
Editing the table does not automatically refit the model. After making changes, click Fit model again to update the results.
Fit diagnostics
Use Grid to view the diagnostic charts together or Single to enlarge one chart. Select a chart using the Chart control:
Fit: shows the observations and fitted curve.
Obs-pred: compares observed and fitted values against a reference line representing exact agreement.
Residuals: shows the differences between the observations and fitted values.
Estimates: shows the parameter estimates and their uncertainty where available.
For secondary models, select the environmental variable for the horizontal axis. When another variable has multiple values, the chart displays a group of curves showing its effect. Use the vertical slider to select the highlighted curve.
The chart displays up to five curves. When there are more than five distinct values, it uses representative values across the data. Other environmental variables are held at their median values when drawing the curves.
Fit statistics and parameter estimates
SSE: the sum of squared differences between observed and fitted values.
RMSE: summarises the size of fitting errors while giving greater weight to larger errors.
MAE: the average absolute difference between observed and fitted values.
R² and adjusted R²: describe how much of the variation in the observations is explained by the model. Adjusted R² accounts for the number of estimated parameters.
AIC and BIC: compare fit quality while accounting for the number of estimated parameters. Lower values indicate a better balance when comparing models fitted to the same observations.
The table shows each estimated parameter’s value and its standard error where available. Standard errors describe the uncertainty in the parameter estimates.
Note
Review the fitted curve, residuals and parameter estimates together. A completed fit means the fitting procedure returned a solution. Its suitability still depends on the data and selected model.
Compare fits
To compare models, fit one model then select another model and click Fit model using the same dataset. Repeat for any additional models you want to compare.
Enable Compare fits once at least two compatible fits are available. The Fit chart overlays their fitted curves and the statistics table shows their results together. The selected model is shown with a solid line and the other models with dashed lines. The Obs-pred, Residuals and Estimates charts show results for the currently selected model.
Use the overlaid curves to compare how closely each model follows the observations. Compare the error statistics alongside AIC and BIC to assess fit quality while accounting for the number of estimated parameters. Compatible fits use the same dataset, model type, process and measured variable.
Save a fitted secondary model
After fitting a secondary model, click Save model in Fit diagnostics. Enter a Model label and select the Organism, Food matrix and Environmental factor. Review the Fitted data range then click Save.
The stored model becomes available in the secondary model list for compatible step modules in the simulations workflow.
Note
Stored models are saved in your browser. Clearing browser data removes them, so download any fitted models you want to retain, transfer or share.
Manage stored models
Click View stored fitted models to open Stored fitted models. Use Search to find a model and its controls to manage it:
Edit name or context: changes the model's label or associated organism, food matrix and environmental factor.
Download fitted model: saves a model file that can be imported later or shared with another user.
Delete fitted model: removes the model from browser storage.
Import a downloaded model through Stored fitted models or the Add module dialog in the simulations workflow.
Model Database
Model Database shows the models available in Temere's expandable database for different organisms, food matrices, environmental factors and processes. Use it to check model availability and view the sources of the supplied coefficient values.
Find available models
Enter a term in Search to find entries by organism, model name, process or source.
Use the filters to narrow the results:
Organism: select the organism. Strain information is included where available.
Matrix: select the food matrix.
Environmental factor: select the inhibiting factor.
Model kind: select primary, secondary or dose-response models.
Process: filter by process where applicable.
Search and filters work together. To broaden the results, clear the search text or return a filter to its All option.
Model information
Models are grouped by organism, food matrix and environmental factor. Each entry shows:
Kind: the model category.
Process: the process described by the model.
Model: the model name.
Sets: the number of available combinations of coefficient values for that model and those conditions.
Sources: the sources associated with those coefficient values.
Click a column heading to sort the results.
Settings and Exports
Simulation settings control the random sampling used by Preview, Simulations and several tools. PDF reports let you save and share the scenario configuration and analysis results.
Seed settings
Click Simulation settings at the bottom of the sidebar. Separate seed settings are available for Preview, Simulations, Dose-response, Sampling Plan and Model Fitting.
A seed determines the sequence of random values used during a calculation:
Random: uses a new seed for each run. Results can differ because the sampled values change.
Fixed: uses the entered Seed value for each run. Keep the same seed, inputs and calculation settings to reproduce the results.
For Fixed, enter a whole number between 1 and 2147483647. Click Done after choosing your settings.
Note
Saving a scenario after a successful simulation also saves the seed used for that run, provided the scenario inputs have not changed.
Loading the scenario restores this seed to the Simulations seed setting but does not automatically select Fixed.
Select Fixed before running the loaded scenario to reproduce the simulation.
Seeds in Preview and Model Fitting
Preview uses a fixed seed by default, keeping its sampled results consistent while you review a scenario. When all inputs are single values, the preview is deterministic and does not depend on random sampling.
In Model Fitting, the seed controls the automatically generated starting points used during fitting. A fixed seed lets you repeat the fit using the same starting points and dataset.
Generate a simulation report
After a successful simulation, click Generate report in Simulations or a simulations tool.
Enter a File name and choose the content to include:
Simulation summary
Charts
Dose-response results
Sensitivity analysis results
Time optimisation results
Sampling plan results
Tool results become available for inclusion after the corresponding analysis has completed for the current simulation. If you run a new simulation, run the relevant tools again before including their updated results.
Click Download to save the report as a PDF. The report includes the scenario configuration and seed information alongside the selected results and charts that may be used in publications provided Temere is acknowledged.
Generate a distribution fitting report
After fitting distributions, click Generate report in Distribution Fitting.
Enter a File name and select the fitted distributions to include. Use Select all or Deselect all to change the selection together then click Download to save the PDF.
The report includes the selected distributions, fitted parameters, fit statistics and diagnostic charts.
Troubleshooting
If a calculation cannot proceed, review the displayed message to identify the affected input or setting. Correct the issue then run the calculation again.
Input errors and validity warnings
Red indicators identify invalid inputs that must be corrected before the calculation can proceed. Check for missing values, invalid distribution settings or values outside the parameter's physical limits.
Sampling error shown in progress bar during simulations
Yellow warnings identify inputs outside the selected models' common valid range. The simulation can still run with these inputs. Open the parameter card's footer to review the individual model ranges and their common zone.
Click View in a validity warning notification to open the affected parameter card.
Warning notification
Sampling and truncation
Sampling can fail when the truncation limits retain only an extremely small part of the selected distribution. This can happen when the limits are far from the values around which the distribution is concentrated.
Review the density chart and check the distribution parameters alongside the truncation limits. Adjust them to represent the intended range of values.
For a fitted distribution selected from a static file, check its parameters and truncation limits as well as the original data.
File inputs
If a file input prevents the simulation from running, open its parameter card and review the highlighted cells and validation message.
Check that the table contains numeric values in the expected columns and that the values remain within the parameter's physical limits. For dynamic profiles, also check the time column and the sequence of time points.
See Parameter Cards for guidance on static files and dynamic profiles.
Simulations tools are unavailable
Run Simulations successfully before using Dose-response, Sampling Plan, Sensitivity Analysis or Time Optimisation. These tools use the latest successful simulation results.
After changing the scenario, run Simulations again then repeat the relevant tool analyses to update their results.
Calculations take longer than expected
Large numbers of simulated products, dynamic file profiles and fitting several distributions increase computation time. Time optimisation also runs repeated calculations while searching for a solution.
Use Preview to check the scenario's general behaviour before running Simulations with a large number of iterations.
Results differ between runs
Random sampling produces differences between runs. To reproduce a calculation, use Fixed seed settings and keep the inputs and calculation settings unchanged.
See Settings and Exports for seed settings and restoring the seed from a saved scenario.
Feedback
Share anonymous feedback about Temere.
Loading feedback form...
Generate distribution fitting report
Choose the fitted distributions to include. The PDF is generated in your browser.
Generate simulation report
Choose the content to include. The PDF is generated in your browser.