ThroughputHigher is desirable
Scenario ANot runpatients in 30 days
Scenario BNot runpatients in 30 days
Client-side stochastic model
WARD CAPACITY AND PATIENT FLOW SIMULATOR
Test your assumptions.
See how the ward responds.
Compare how open beds, room configuration, unscheduled demand and clinical placement constraints affect admissions, occupancy and rejection.
AIM
Understand the tradeoffs behind ward capacity.
This tool explores how physical bed count, room mix, non-elective demand variability and clinical placement constraints together shape achievable patient throughput and occupancy. Potential emergency and intensive care admission requests cannot be scheduled. Ward fullness may optionally suppress some potential demand before a genuine request is made. Each genuine request is then either granted, meaning the patient is admitted to this ward, or rejected because the patient cannot be placed. A rejected request means the patient remains in emergency or ICU, or is transferred to another ward or hospital, depending on the local context. The tool supports decisions about how many beds to operate, but does not model nursing or medical staffing capacity.
This tool does not model staffing requirements, detailed costs, or clinical acuity. All input values are placeholder defaults, not measured data from any specific ward, and none should be treated as more reliable than any other. The simulation model itself, including its structure and assumptions, has not been validated against real-world ward data. Under high demand, reported occupancy is driven largely by the demand suppression curve assumption rather than by ward capacity alone when that mechanism is enabled. The model assumes no day-of-week or seasonal demand pattern, no correlation between patient factors, and no interaction with other wards or hospitals. Suppressed potential demand and rejected admission requests are counted, but their downstream consequences are not modeled. Estimated DRG revenue represents gross case-based revenue only. Any net estimated result, when shown, subtracts only the user-assumed staffed bed cost and includes no other cost category, such as equipment, medication, administration or overhead; it must not be read as a complete or accurate net financial outcome or profitability estimate for either scenario. For all of these reasons together, no result from this tool should be used as the basis for any real operational or staffing decision.
01 · CONFIGURE
Compare two operating scenarios
PARAMETER SET
Scenario A
38physical beds
38beds open
8 + 15×2room mix
Ward structure
Demand and length of stay
Cleaning, demand suppression and waiting
PARAMETER SET
Scenario B
38physical beds
34beds open
8 + 15×2room mix
Ward structure
Demand and length of stay
Cleaning, demand suppression and waiting
Ready
FIXED 30 DAY RESULTS
Scenario comparison
Mean ± one standard deviation across independent replications
Average occupancyContext dependent
Scenario ANot runbeds and % of active staffed beds
Scenario BNot runbeds and % of active staffed beds
Rejection rateHigher is undesirable
Scenario ANot runrejected patients and % of genuine admission requests
Scenario BNot runrejected patients and % of genuine admission requests
Estimated DRG revenueEstimated revenue only
Scenario ANot run30 day estimateNot runAnnualized projection (30 days × 12)
Scenario BNot run30 day estimateNot runAnnualized projection (30 days × 12)
Scenario A minus Scenario BNot run30 daysNot runAnnualized projection (30 days × 12)
Secondary outcomes
Beds closed for single-room usebeds and % of active staffed beds
Average boarding timehours among admitted patients
Suppressed demandsuppressed potential requests and % of potential demand
Rejected: no compatible bedrejected patients and % of genuine admission requests
Rejected: no eligible single-room-capable bedrejected patients and % of genuine admission requests
Scenario A
Not runbeds and % of active staffed beds
Not runhours among admitted patients
Not runsuppressed potential requests and % of potential demand
Not runrejected patients and % of genuine admission requests
Not runrejected patients and % of genuine admission requests
Scenario B
Not runbeds and % of active staffed beds
Not runhours among admitted patients
Not runsuppressed potential requests and % of potential demand
Not runrejected patients and % of genuine admission requests
Not runrejected patients and % of genuine admission requests
Run the scenarios to see granted admission requests on each of the 30 analysis days.
Run the scenarios to see occupancy on each of the 30 analysis days.
02 · PRIMARY ANALYSIS
Trade-off explorer
All other inputs stay fixed at Scenario A; every point uses a 30 day analysis window.
ThroughputMean curve with ±1 SD; hover or tap a point for exact values
patients in 30 daysRun the Trade-off explorer to see the replicated response curve.
03 · METHODS
A discrete-event simulation of unscheduled ward demand
Each replication begins with the selected warm-up period, followed by a fixed 30 day analysis window. Potential request events follow a Gamma–Poisson process and patient length of stay follows a right-skewed log-normal distribution. A scenario may keep a fixed staffed bed count, or use an elastic concept that activates reserve physical beds when single-room conversions close usable capacity and returns them to standby when no longer needed, up to the scenario's total physical beds. When demand suppression is enabled, every non-empty active staffed bed contributes to current fullness, with the current active staffed bed count as denominator. A logistic survival probability may suppress a potential event before it becomes a genuine admission request; suppressed events receive no patient characteristics and are not included in rejection rates. When suppression is disabled, every potential event becomes a genuine request. Genuine requests are assessed using room type, gender matching and the scenario's user-configured chance of requiring a single room. If no suitable bed exists, the request may wait only for the configured maximum wait before rejection; zero means immediate rejection. Boarding time runs from submission of the admission request to actual admission for granted requests, including zero for immediate admissions; rejected requests are excluded. Estimated financial revenue is calculated as each scenario's simulated patient throughput multiplied by an assumed average Swiss DRG return per patient, itself calculated as the shared baserate multiplied by the shared case mix index. The same ward-wide average return is applied to every discharged patient. Staffed bed cost can optionally be included; within each replication it is calculated as the scenario's time-weighted average number of active staffed beds multiplied by the shared daily bed cost and by 30, regardless of occupancy. The net estimated result is revenue minus this staffed bed cost. Annualized financial values are the corresponding 30 day results multiplied by twelve rather than separately simulated years. These remain simplified estimates and do not represent a full financial or cost analysis. All results use only the 30 analysis days and show means and standard deviations across replications. A fixed seed reproduces an unchanged scenario exactly, but different parameter sets can consume random numbers in different orders, so replicated averages remain the reliable basis for comparing scenarios.
04 · HOW THIS WORKS
A deliberately transparent flow model
Potential requests vary by day
A Gamma–Poisson mixture creates adjustable daily burstiness in potential admission demand. Candidate times are random within each day; emergency and ICU demand is never scheduled by the ward.
Fullness can suppress demand
When enabled, all non-empty active staffed beds, including cleaning beds and beds closed by conversion to single use, enter the logistic demand suppression calculation. Some potential events never become genuine admission requests and are reported separately as suppressed demand. When disabled, every potential event proceeds to placement. Standby beds do not enter the calculation.
Placement follows room rules
Regular patients always complete an occupied same-gender double room first. When opening a fresh room, they follow the scenario's double-first or single-first priority and preserve configured whole-room reserve floors while an unreserved option remains. Reserved rooms become available as a last resort. Patients with a single room need continue to use the separate single-capable placement rules.
A single room need can close a bed
At the user-configured rate, some patients submitting admission requests cannot share a room because of infectious isolation, delirium, terminal illness with expected imminent death, or similar patient factors. If granted, they use a single room or convert an eligible empty double room to single use. When a single room becomes free after cleaning, waiting, not-yet-granted requests retain priority; otherwise the longest-admitted eligible patient is transferred from a converted double room.
Waiting has a hard deadline
The maximum wait before rejection sets how many hours an admission request without a compatible bed may wait. It is granted when a suitable bed opens or rejected exactly at its individual deadline. A timed-out request remains classified by the original placement constraint: regular-bed compatibility or single-room capability. A patient whose request is rejected does not enter this ward.
Capacity can be fixed or elastic
The fixed system leaves conversion closures uncompensated. The optional elastic concept opens reserve physical beds to maintain its usable target, then returns empty elastic beds to standby when the extra capacity is no longer needed.
Throughput and staffed beds shape finances
The shared baserate multiplied by the case mix index gives one assumed DRG return per discharged patient. When staffed bed cost is enabled, each scenario's actual time-weighted active staffed beds are charged the shared daily cost regardless of occupancy, and this cost is subtracted from revenue to show a simplified net estimated result. Annualized values are 30 day projections multiplied by twelve.